This is an archive of the discontinued LLVM Phabricator instance.

Differential D105297

[OPENMP]Fix PR50347: Mapping of global scope deep object fails.
ClosedPublic

Authored by ABataev on Jul 1 2021, 10:35 AM.

Details

Reviewers
jdoerfert
mikerice
Commits
rGd04d9220e1c3: [OPENMP]Fix PR50347: Mapping of global scope deep object fails.
rG638938117aea: [OPENMP]Fix PR50347: Mapping of global scope deep object fails.
Summary

Changed the we handle llvm::Constants in sizes arrays. ConstExprs and
GlobalValues cannot be used as initializers, need to put them at the
runtime, otherwise there wight be the compilation errors.

Diff Detail

Unit TestsFailed

TimeTest
70 msx64 debian > LLVM.Bindings/Go::go.test

Event Timeline

ABataev created this revision.Jul 1 2021, 10:35 AM
Herald added subscribers: guansong, yaxunl. · View Herald TranscriptJul 1 2021, 10:35 AM
ABataev requested review of this revision.Jul 1 2021, 10:35 AM
Herald added a project: Restricted Project. · View Herald TranscriptJul 1 2021, 10:35 AM
Herald added a subscriber: sstefan1. · View Herald Transcript
Harbormaster completed remote builds in B112030: Diff 355937.Jul 1 2021, 11:49 AM
ABataev updated this revision to Diff 356752.Jul 6 2021, 9:13 AM

Rebase

Harbormaster completed remote builds in B112638: Diff 356752.Jul 6 2021, 10:12 AM
ABataev updated this revision to Diff 357612.Jul 9 2021, 1:10 PM

Rebase

ABataev updated this revision to Diff 360563.Jul 21 2021, 12:46 PM

Rebase

Harbormaster completed remote builds in B115395: Diff 360563.Jul 21 2021, 2:45 PM
ABataev added a comment.Jul 22 2021, 4:36 AM

Better to fix this before the next release.

ABataev updated this revision to Diff 368426.Aug 24 2021, 12:21 PM

Rebase + update

Herald added a subscriber: jfb. · View Herald TranscriptAug 24 2021, 12:21 PM
Harbormaster completed remote builds in B121023: Diff 368426.Aug 24 2021, 1:51 PM
ABataev updated this revision to Diff 376945.Oct 4 2021, 10:42 AM

Rebase

Harbormaster completed remote builds in B126859: Diff 376945.Oct 4 2021, 12:04 PM
ABataev updated this revision to Diff 396531.Dec 29 2021, 7:10 AM

Rebase

Harbormaster completed remote builds in B140900: Diff 396531.Dec 29 2021, 7:40 AM
ABataev updated this revision to Diff 406099.Feb 4 2022, 1:54 PM

Rebase

Harbormaster completed remote builds in B147714: Diff 406099.Feb 4 2022, 3:18 PM
This revision was not accepted when it landed; it landed in state Needs Review.Feb 24 2022, 11:55 AM
This revision was landed with ongoing or failed builds.
Closed by commit rG638938117aea: [OPENMP]Fix PR50347: Mapping of global scope deep object fails. (authored by ABataev). · Explain Why
This revision was automatically updated to reflect the committed changes.
ABataev added a commit: rG638938117aea: [OPENMP]Fix PR50347: Mapping of global scope deep object fails..
ABataev added a reverting change: rGca6fa71b7ee1: Revert "[OPENMP]Fix PR50347: Mapping of global scope deep object fails.".Feb 24 2022, 12:05 PM
ABataev added a commit: rGd04d9220e1c3: [OPENMP]Fix PR50347: Mapping of global scope deep object fails..Feb 25 2022, 10:58 AM

Revision Contents

PathSize
clang/
lib/
CodeGen/
47 lines
test/
OpenMP/
2009 lines
67 lines
27 lines
402 lines
3 lines
19 lines
12 lines
17 lines
3 lines
17 lines
3 lines
42 lines
12 lines
13 lines
26 lines
100 lines
20 lines
78 lines
29 lines
13 lines
20 lines
11 lines
32 lines
24 lines
862 lines
20 lines
2248 lines
2392 lines
2488 lines
90 lines
2968 lines
2392 lines
484 lines
1436 lines
484 lines
1500 lines
5472 lines
484 lines
1500 lines
5472 lines
2312 lines
484 lines
1436 lines
60 lines
3 lines
624 lines
485 lines
1320 lines
624 lines
485 lines
1460 lines
5360 lines
768 lines
1036 lines
1460 lines
5360 lines
1492 lines
820 lines
1320 lines
1284 lines

Diff 396531

clang/lib/CodeGen/CGOpenMPRuntime.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show All 23 Lines
#include "clang/AST/StmtVisitor.h" #include "clang/AST/StmtVisitor.h"
#include "clang/Basic/BitmaskEnum.h" #include "clang/Basic/BitmaskEnum.h"
#include "clang/Basic/FileManager.h" #include "clang/Basic/FileManager.h"
#include "clang/Basic/OpenMPKinds.h" #include "clang/Basic/OpenMPKinds.h"
#include "clang/Basic/SourceManager.h" #include "clang/Basic/SourceManager.h"
#include "clang/CodeGen/ConstantInitBuilder.h" #include "clang/CodeGen/ConstantInitBuilder.h"
#include "llvm/ADT/ArrayRef.h" #include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/SetOperations.h" #include "llvm/ADT/SetOperations.h"
#include "llvm/ADT/SmallBitVector.h"
#include "llvm/ADT/StringExtras.h" #include "llvm/ADT/StringExtras.h"
#include "llvm/Bitcode/BitcodeReader.h" #include "llvm/Bitcode/BitcodeReader.h"
#include "llvm/IR/Constants.h" #include "llvm/IR/Constants.h"
#include "llvm/IR/DerivedTypes.h" #include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/GlobalValue.h" #include "llvm/IR/GlobalValue.h"
#include "llvm/IR/Value.h" #include "llvm/IR/Value.h"
#include "llvm/Support/AtomicOrdering.h" #include "llvm/Support/AtomicOrdering.h"
#include "llvm/Support/Format.h" #include "llvm/Support/Format.h"
▲ Show 20 Lines • Show All 9,531 Lines • ▼ Show 20 Linesstatic void emitOffloadingArrays(
// Reset the array information. // Reset the array information.
Info.clearArrayInfo(); Info.clearArrayInfo();
Info.NumberOfPtrs = CombinedInfo.BasePointers.size(); Info.NumberOfPtrs = CombinedInfo.BasePointers.size();
if (Info.NumberOfPtrs) { if (Info.NumberOfPtrs) {
// Detect if we have any capture size requiring runtime evaluation of the // Detect if we have any capture size requiring runtime evaluation of the
// size so that a constant array could be eventually used. // size so that a constant array could be eventually used.
bool hasRuntimeEvaluationCaptureSize = false;
for (llvm::Value *S : CombinedInfo.Sizes)
if (!isa<llvm::Constant>(S)) {
hasRuntimeEvaluationCaptureSize = true;
break;
}
llvm::APInt PointerNumAP(32, Info.NumberOfPtrs, /*isSigned=*/true); llvm::APInt PointerNumAP(32, Info.NumberOfPtrs, /*isSigned=*/true);
QualType PointerArrayType = Ctx.getConstantArrayType( QualType PointerArrayType = Ctx.getConstantArrayType(
Ctx.VoidPtrTy, PointerNumAP, nullptr, ArrayType::Normal, Ctx.VoidPtrTy, PointerNumAP, nullptr, ArrayType::Normal,
/*IndexTypeQuals=*/0); /*IndexTypeQuals=*/0);
Info.BasePointersArray = Info.BasePointersArray =
CGF.CreateMemTemp(PointerArrayType, ".offload_baseptrs").getPointer(); CGF.CreateMemTemp(PointerArrayType, ".offload_baseptrs").getPointer();
Info.PointersArray = Info.PointersArray =
CGF.CreateMemTemp(PointerArrayType, ".offload_ptrs").getPointer(); CGF.CreateMemTemp(PointerArrayType, ".offload_ptrs").getPointer();
Address MappersArray = Address MappersArray =
CGF.CreateMemTemp(PointerArrayType, ".offload_mappers"); CGF.CreateMemTemp(PointerArrayType, ".offload_mappers");
Info.MappersArray = MappersArray.getPointer(); Info.MappersArray = MappersArray.getPointer();
// If we don't have any VLA types or other types that require runtime // If we don't have any VLA types or other types that require runtime
// evaluation, we can use a constant array for the map sizes, otherwise we // evaluation, we can use a constant array for the map sizes, otherwise we
// need to fill up the arrays as we do for the pointers. // need to fill up the arrays as we do for the pointers.
QualType Int64Ty = QualType Int64Ty =
Ctx.getIntTypeForBitwidth(/*DestWidth=*/64, /*Signed=*/1); Ctx.getIntTypeForBitwidth(/*DestWidth=*/64, /*Signed=*/1);
if (hasRuntimeEvaluationCaptureSize) { SmallVector<llvm::Constant *> ConstSizes;
QualType SizeArrayType = Ctx.getConstantArrayType( llvm::SmallBitVector RuntimeSizes(CombinedInfo.Sizes.size());
Int64Ty, PointerNumAP, nullptr, ArrayType::Normal,
/*IndexTypeQuals=*/0);
Info.SizesArray =
CGF.CreateMemTemp(SizeArrayType, ".offload_sizes").getPointer();
} else {
// We expect all the sizes to be constant, so we collect them to create
// a constant array.
SmallVector<llvm::Constant *, 16> ConstSizes;
for (unsigned I = 0, E = CombinedInfo.Sizes.size(); I < E; ++I) { for (unsigned I = 0, E = CombinedInfo.Sizes.size(); I < E; ++I) {
if (IsNonContiguous && if (IsNonContiguous &&
(CombinedInfo.Types[I] & MappableExprsHandler::OMP_MAP_NON_CONTIG)) { (CombinedInfo.Types[I] & MappableExprsHandler::OMP_MAP_NON_CONTIG)) {
ConstSizes.push_back(llvm::ConstantInt::get( ConstSizes.push_back(llvm::ConstantInt::get(
CGF.Int64Ty, CombinedInfo.NonContigInfo.Dims[I])); CGF.Int64Ty, CombinedInfo.NonContigInfo.Dims[I]));
} else { } else {
if (auto *CI = dyn_cast<llvm::Constant>(CombinedInfo.Sizes[I]))
if (!isa<llvm::ConstantExpr>(CI) && !isa<llvm::GlobalValue>(CI)) {
ConstSizes.push_back(cast<llvm::Constant>(CombinedInfo.Sizes[I])); ConstSizes.push_back(cast<llvm::Constant>(CombinedInfo.Sizes[I]));
continue;
}
ConstSizes.push_back(llvm::ConstantInt::get(CGF.Int64Ty, 0));
RuntimeSizes.set(I);
} }
} }
if (RuntimeSizes.all()) {
QualType SizeArrayType = Ctx.getConstantArrayType(
Int64Ty, PointerNumAP, nullptr, ArrayType::Normal,
/*IndexTypeQuals=*/0);
Info.SizesArray =
CGF.CreateMemTemp(SizeArrayType, ".offload_sizes").getPointer();
} else {
auto *SizesArrayInit = llvm::ConstantArray::get( auto *SizesArrayInit = llvm::ConstantArray::get(
llvm::ArrayType::get(CGM.Int64Ty, ConstSizes.size()), ConstSizes); llvm::ArrayType::get(CGM.Int64Ty, ConstSizes.size()), ConstSizes);
std::string Name = CGM.getOpenMPRuntime().getName({"offload_sizes"}); std::string Name = CGM.getOpenMPRuntime().getName({"offload_sizes"});
auto *SizesArrayGbl = new llvm::GlobalVariable( auto *SizesArrayGbl = new llvm::GlobalVariable(
CGM.getModule(), SizesArrayInit->getType(), CGM.getModule(), SizesArrayInit->getType(), !RuntimeSizes.any(),
/*isConstant=*/true, llvm::GlobalValue::PrivateLinkage, llvm::GlobalValue::PrivateLinkage, SizesArrayInit, Name);
SizesArrayInit, Name);
SizesArrayGbl->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global); SizesArrayGbl->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
Info.SizesArray = SizesArrayGbl; Info.SizesArray = SizesArrayGbl;
} }
// The map types are always constant so we don't need to generate code to // The map types are always constant so we don't need to generate code to
// fill arrays. Instead, we create an array constant. // fill arrays. Instead, we create an array constant.
SmallVector<uint64_t, 4> Mapping(CombinedInfo.Types.size(), 0); SmallVector<uint64_t, 4> Mapping(CombinedInfo.Types.size(), 0);
llvm::copy(CombinedInfo.Types, Mapping.begin()); llvm::copy(CombinedInfo.Types, Mapping.begin());
▲ Show 20 Lines • Show All 57 Lines • ▼ Show 20 Lines for (unsigned I = 0; I < Info.NumberOfPtrs; ++I) {
llvm::Value *P = CGF.Builder.CreateConstInBoundsGEP2_32( llvm::Value *P = CGF.Builder.CreateConstInBoundsGEP2_32(
llvm::ArrayType::get(CGM.VoidPtrTy, Info.NumberOfPtrs), llvm::ArrayType::get(CGM.VoidPtrTy, Info.NumberOfPtrs),
Info.PointersArray, 0, I); Info.PointersArray, 0, I);
P = CGF.Builder.CreatePointerBitCastOrAddrSpaceCast( P = CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(
P, PVal->getType()->getPointerTo(/*AddrSpace=*/0)); P, PVal->getType()->getPointerTo(/*AddrSpace=*/0));
Address PAddr(P, Ctx.getTypeAlignInChars(Ctx.VoidPtrTy)); Address PAddr(P, Ctx.getTypeAlignInChars(Ctx.VoidPtrTy));
CGF.Builder.CreateStore(PVal, PAddr); CGF.Builder.CreateStore(PVal, PAddr);
if (hasRuntimeEvaluationCaptureSize) { if (RuntimeSizes.test(I)) {
llvm::Value *S = CGF.Builder.CreateConstInBoundsGEP2_32( llvm::Value *S = CGF.Builder.CreateConstInBoundsGEP2_32(
llvm::ArrayType::get(CGM.Int64Ty, Info.NumberOfPtrs), llvm::ArrayType::get(CGM.Int64Ty, Info.NumberOfPtrs),
Info.SizesArray, Info.SizesArray,
/*Idx0=*/0, /*Idx0=*/0,
/*Idx1=*/I); /*Idx1=*/I);
Address SAddr(S, Ctx.getTypeAlignInChars(Int64Ty)); Address SAddr(S, Ctx.getTypeAlignInChars(Int64Ty));
CGF.Builder.CreateStore(CGF.Builder.CreateIntCast(CombinedInfo.Sizes[I], CGF.Builder.CreateStore(CGF.Builder.CreateIntCast(CombinedInfo.Sizes[I],
CGM.Int64Ty, CGM.Int64Ty,
▲ Show 20 Lines • Show All 3,442 Lines • Show Last 20 Lines

clang/test/OpenMP/reduction_implicit_map.cpp

// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[ .].+[.|,]" --prefix-filecheck-ir-name _
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-cuda-mode -x c++ \ // RUN: %clang_cc1 -verify -fopenmp -fopenmp-cuda-mode -x c++ \
// RUN: -triple powerpc64le-unknown-unknown -DCUDA \ // RUN: -triple powerpc64le-unknown-unknown -DCUDA \
// RUN: -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o \ // RUN: -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o \
// RUN: %t-ppc-host.bc // RUN: %t-ppc-host.bc
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-cuda-mode -x c++ \ // RUN: %clang_cc1 -verify -fopenmp -fopenmp-cuda-mode -x c++ \
// RUN: -triple nvptx64-unknown-unknown -DCUA \ // RUN: -triple nvptx64-unknown-unknown -DCUA \
// RUN: -fopenmp-targets=nvptx64-nvidia-cuda -DCUDA -emit-llvm %s \ // RUN: -fopenmp-targets=nvptx64-nvidia-cuda -DCUDA -emit-llvm %s \
Show All 20 Linesint foo(int n) {
double *e; double *e;
//no error and no implicit map generated for e[:1] //no error and no implicit map generated for e[:1]
#pragma omp target parallel reduction(+: e[:1]) #pragma omp target parallel reduction(+: e[:1])
*e=10; *e=10;
; ;
return 0; return 0;
} }
// CHECK-NOT @.offload_maptypes // CHECK-NOT @.offload_maptypes
// CHECK: call void @__kmpc_nvptx_end_reduce_nowait(
#elif defined(DIAG) #elif defined(DIAG)
class S2 { class S2 {
mutable int a; mutable int a;
public: public:
S2():a(0) { } S2():a(0) { }
S2(S2 &s2):a(s2.a) { } S2(S2 &s2):a(s2.a) { }
S2 &operator +(S2 &s); S2 &operator +(S2 &s);
}; };
int bar() { int bar() {
S2 o[5]; S2 o[5];
//warnig "copyable and not guaranteed to be mapped correctly" and //warnig "copyable and not guaranteed to be mapped correctly" and
//implicit map generated. //implicit map generated.
#pragma omp target parallel reduction(+:o[0]) //expected-warning {{Type 'S2' is not trivially copyable and not guaranteed to be mapped correctly}} #pragma omp target parallel reduction(+:o[0]) //expected-warning {{Type 'S2' is not trivially copyable and not guaranteed to be mapped correctly}}
for (int i = 0; i < 10; i++); for (int i = 0; i < 10; i++);
double b[10][10][10]; double b[10][10][10];
//no error no implicit map generated, the map for b is generated but not //no error no implicit map generated, the map for b is generated but not
//for b[0:2][2:4][1]. //for b[0:2][2:4][1].
#pragma omp target parallel for reduction(task, +: b[0:2][2:4][1]) #pragma omp target parallel for reduction(task, +: b[0:2][2:4][1])
for (long long i = 0; i < 10; ++i); for (long long i = 0; i < 10; ++i);
return 0; return 0;
} }
// map for variable o // map for variable o
// CHECK1: offload_sizes = private unnamed_addr constant [1 x i64] [i64 4]
// CHECK1: offload_maptypes = private unnamed_addr constant [1 x i64] [i64 547]
// map for b: // map for b:
// CHECK1: @.offload_sizes{{.*}} = private unnamed_addr constant [1 x i64] [i64 8000]
// CHECK1: @.offload_maptypes{{.*}} = private unnamed_addr constant [1 x i64] [i64 547]
#else #else
// expected-no-diagnostics // expected-no-diagnostics
// generate implicit map for array elements or array sections in reduction // generate implicit map for array elements or array sections in reduction
// clause. In following case: the implicit map is generate for output[0] // clause. In following case: the implicit map is generate for output[0]
// with map size 4 and output[:3] with map size 12. // with map size 4 and output[:3] with map size 12.
void sum(int* input, int size, int* output) void sum(int* input, int size, int* output)
{ {
#pragma omp target teams distribute parallel for reduction(+: output[0]) \ #pragma omp target teams distribute parallel for reduction(+: output[0]) \
map(to: input [0:size]) map(to: input [0:size])
for (int i = 0; i < size; i++) for (int i = 0; i < size; i++)
output[0] += input[i]; output[0] += input[i];
#pragma omp target teams distribute parallel for reduction(+: output[:3]) \ #pragma omp target teams distribute parallel for reduction(+: output[:3]) \
map(to: input [0:size]) map(to: input [0:size])
for (int i = 0; i < size; i++) for (int i = 0; i < size; i++)
output[0] += input[i]; output[0] += input[i];
int a[10]; int a[10];
#pragma omp target parallel reduction(+: a[:2]) #pragma omp target parallel reduction(+: a[:2])
for (int i = 0; i < size; i++) for (int i = 0; i < size; i++)
; ;
#pragma omp target parallel reduction(+: a[3]) #pragma omp target parallel reduction(+: a[3])
for (int i = 0; i < size; i++) for (int i = 0; i < size; i++)
; ;
} }
//CHECK2: @.offload_sizes = private unnamed_addr constant [2 x i64] [i64 4, i64 8]
//CHECK2: @.offload_maptypes.10 = private unnamed_addr constant [2 x i64] [i64 800, i64 547]
//CHECK2: @.offload_sizes.13 = private unnamed_addr constant [2 x i64] [i64 4, i64 4]
//CHECK2: @.offload_maptypes.14 = private unnamed_addr constant [2 x i64] [i64 800, i64 547]
//CHECK2: define dso_local void @_Z3sumPiiS_
//CHECK2-NEXT: entry
//CHECK2-NEXT: [[INP:%.*]] = alloca i32*
//CHECK2-NEXT: [[SIZE:%.*]] = alloca i32
//CHECK2-NEXT: [[OUTP:%.*]] = alloca i32*
//CHECK2: [[OFFSIZE:%.*]] = alloca [3 x i64]
//CHECK2: [[OFFSIZE10:%.*]] = alloca [3 x i64]
//CHECK2: [[T15:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[OFFSIZE]], i32 0, i32 0
//CHECK2-NEXT: store i64 4, i64* [[T15]]
//CHECK2: [[T21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[OFFSIZE]], i32 0, i32 1
//CHECK2-NEXT: store i64 4, i64* [[T21]]
//CHECK2: [[T53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[OFFSIZE10]], i32 0, i32 0
//CHECK2-NEXT: store i64 4, i64* [[T53]]
//CHECK2: [[T59:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[OFFSIZE10]], i32 0, i32 1
//CHECK2-NEXT: store i64 12, i64* [[T59]]
#endif #endif
int main() int main()
{ {
#if defined(CUDA) #if defined(CUDA)
int a = foo(10); int a = foo(10);
#elif defined(DIAG) #elif defined(DIAG)
int a = bar(); int a = bar();
#else #else
const int size = 100; const int size = 100;
int *array = new int[size]; int *array = new int[size];
int result = 0; int result = 0;
sum(array, size, &result); sum(array, size, &result);
#endif #endif
return 0; return 0;
} }
// CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l32
// CHECK-SAME: (double* [[E:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK-NEXT: entry:
// CHECK-NEXT: [[E_ADDR:%.*]] = alloca double*, align 8
// CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x i8*], align 8
// CHECK-NEXT: store double* [[E]], double** [[E_ADDR]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true)
// CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK: user_code.entry:
// CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK-NEXT: [[TMP2:%.*]] = load double*, double** [[E_ADDR]], align 8
// CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-NEXT: [[TMP4:%.*]] = bitcast double* [[TMP2]] to i8*
// CHECK-NEXT: store i8* [[TMP4]], i8** [[TMP3]], align 8
// CHECK-NEXT: [[TMP5:%.*]] = bitcast [1 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
// CHECK-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB2]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, double*)* @__omp_outlined__ to i8*), i8* null, i8** [[TMP5]], i64 1)
// CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 true)
// CHECK-NEXT: ret void
// CHECK: worker.exit:
// CHECK-NEXT: ret void
//
//
// CHECK-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], double* [[E:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK-NEXT: entry:
// CHECK-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK-NEXT: [[E_ADDR:%.*]] = alloca double*, align 8
// CHECK-NEXT: [[E2:%.*]] = alloca double, align 8
// CHECK-NEXT: [[TMP:%.*]] = alloca double*, align 8
// CHECK-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK-NEXT: store double* [[E]], double** [[E_ADDR]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load double*, double** [[E_ADDR]], align 8
// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[TMP0]], i64 0
// CHECK-NEXT: [[TMP1:%.*]] = load double*, double** [[E_ADDR]], align 8
// CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds double, double* [[TMP1]], i64 0
// CHECK-NEXT: store double 0.000000e+00, double* [[E2]], align 8
// CHECK-NEXT: [[TMP2:%.*]] = load double*, double** [[E_ADDR]], align 8
// CHECK-NEXT: [[TMP3:%.*]] = ptrtoint double* [[TMP2]] to i64
// CHECK-NEXT: [[TMP4:%.*]] = ptrtoint double* [[ARRAYIDX]] to i64
// CHECK-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (double* getelementptr (double, double* null, i32 1) to i64)
// CHECK-NEXT: [[TMP7:%.*]] = getelementptr double, double* [[E2]], i64 [[TMP6]]
// CHECK-NEXT: store double* [[TMP7]], double** [[TMP]], align 8
// CHECK-NEXT: [[TMP8:%.*]] = load double*, double** [[TMP]], align 8
// CHECK-NEXT: store double 1.000000e+01, double* [[TMP8]], align 8
// CHECK-NEXT: [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
// CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK-NEXT: [[TMP12:%.*]] = bitcast double* [[E2]] to i8*
// CHECK-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8
// CHECK-NEXT: [[TMP13:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK-NEXT: [[TMP14:%.*]] = call i32 @__kmpc_nvptx_parallel_reduce_nowait_v2(%struct.ident_t* @[[GLOB2]], i32 [[TMP10]], i32 1, i64 8, i8* [[TMP13]], void (i8*, i16, i16, i16)* @_omp_reduction_shuffle_and_reduce_func, void (i8*, i32)* @_omp_reduction_inter_warp_copy_func)
// CHECK-NEXT: [[TMP15:%.*]] = icmp eq i32 [[TMP14]], 1
// CHECK-NEXT: br i1 [[TMP15]], label [[DOTOMP_REDUCTION_THEN:%.*]], label [[DOTOMP_REDUCTION_DONE:%.*]]
// CHECK: .omp.reduction.then:
// CHECK-NEXT: [[TMP16:%.*]] = load double, double* [[ARRAYIDX]], align 8
// CHECK-NEXT: [[TMP17:%.*]] = load double, double* [[E2]], align 8
// CHECK-NEXT: [[ADD:%.*]] = fadd double [[TMP16]], [[TMP17]]
// CHECK-NEXT: store double [[ADD]], double* [[ARRAYIDX]], align 8
// CHECK-NEXT: call void @__kmpc_nvptx_end_reduce_nowait(i32 [[TMP10]])
// CHECK-NEXT: br label [[DOTOMP_REDUCTION_DONE]]
// CHECK: .omp.reduction.done:
// CHECK-NEXT: ret void
//
//
// CHECK-LABEL: define {{[^@]+}}@_omp_reduction_shuffle_and_reduce_func
// CHECK-SAME: (i8* [[TMP0:%.*]], i16 signext [[TMP1:%.*]], i16 signext [[TMP2:%.*]], i16 signext [[TMP3:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK-NEXT: entry:
// CHECK-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK-NEXT: [[DOTADDR1:%.*]] = alloca i16, align 2
// CHECK-NEXT: [[DOTADDR2:%.*]] = alloca i16, align 2
// CHECK-NEXT: [[DOTADDR3:%.*]] = alloca i16, align 2
// CHECK-NEXT: [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK-NEXT: [[DOTOMP_REDUCTION_ELEMENT:%.*]] = alloca double, align 8
// CHECK-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK-NEXT: store i16 [[TMP1]], i16* [[DOTADDR1]], align 2
// CHECK-NEXT: store i16 [[TMP2]], i16* [[DOTADDR2]], align 2
// CHECK-NEXT: store i16 [[TMP3]], i16* [[DOTADDR3]], align 2
// CHECK-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK-NEXT: [[TMP6:%.*]] = load i16, i16* [[DOTADDR1]], align 2
// CHECK-NEXT: [[TMP7:%.*]] = load i16, i16* [[DOTADDR2]], align 2
// CHECK-NEXT: [[TMP8:%.*]] = load i16, i16* [[DOTADDR3]], align 2
// CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST]], i64 0, i64 0
// CHECK-NEXT: [[TMP12:%.*]] = bitcast i8* [[TMP10]] to double*
// CHECK-NEXT: [[TMP13:%.*]] = getelementptr double, double* [[TMP12]], i64 1
// CHECK-NEXT: [[TMP14:%.*]] = bitcast double* [[TMP13]] to i8*
// CHECK-NEXT: [[TMP15:%.*]] = bitcast double* [[TMP12]] to i64*
// CHECK-NEXT: [[TMP16:%.*]] = bitcast double* [[DOTOMP_REDUCTION_ELEMENT]] to i64*
// CHECK-NEXT: [[TMP17:%.*]] = load i64, i64* [[TMP15]], align 8
// CHECK-NEXT: [[NVPTX_WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
// CHECK-NEXT: [[TMP18:%.*]] = trunc i32 [[NVPTX_WARP_SIZE]] to i16
// CHECK-NEXT: [[TMP19:%.*]] = call i64 @__kmpc_shuffle_int64(i64 [[TMP17]], i16 [[TMP7]], i16 [[TMP18]])
// CHECK-NEXT: store i64 [[TMP19]], i64* [[TMP16]], align 8
// CHECK-NEXT: [[TMP20:%.*]] = getelementptr i64, i64* [[TMP15]], i64 1
// CHECK-NEXT: [[TMP21:%.*]] = getelementptr i64, i64* [[TMP16]], i64 1
// CHECK-NEXT: [[TMP22:%.*]] = bitcast double* [[DOTOMP_REDUCTION_ELEMENT]] to i8*
// CHECK-NEXT: store i8* [[TMP22]], i8** [[TMP11]], align 8
// CHECK-NEXT: [[TMP23:%.*]] = icmp eq i16 [[TMP8]], 0
// CHECK-NEXT: [[TMP24:%.*]] = icmp eq i16 [[TMP8]], 1
// CHECK-NEXT: [[TMP25:%.*]] = icmp ult i16 [[TMP6]], [[TMP7]]
// CHECK-NEXT: [[TMP26:%.*]] = and i1 [[TMP24]], [[TMP25]]
// CHECK-NEXT: [[TMP27:%.*]] = icmp eq i16 [[TMP8]], 2
// CHECK-NEXT: [[TMP28:%.*]] = and i16 [[TMP6]], 1
// CHECK-NEXT: [[TMP29:%.*]] = icmp eq i16 [[TMP28]], 0
// CHECK-NEXT: [[TMP30:%.*]] = and i1 [[TMP27]], [[TMP29]]
// CHECK-NEXT: [[TMP31:%.*]] = icmp sgt i16 [[TMP7]], 0
// CHECK-NEXT: [[TMP32:%.*]] = and i1 [[TMP30]], [[TMP31]]
// CHECK-NEXT: [[TMP33:%.*]] = or i1 [[TMP23]], [[TMP26]]
// CHECK-NEXT: [[TMP34:%.*]] = or i1 [[TMP33]], [[TMP32]]
// CHECK-NEXT: br i1 [[TMP34]], label [[THEN:%.*]], label [[ELSE:%.*]]
// CHECK: then:
// CHECK-NEXT: [[TMP35:%.*]] = bitcast [1 x i8*]* [[TMP5]] to i8*
// CHECK-NEXT: [[TMP36:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST]] to i8*
// CHECK-NEXT: call void @"_omp$reduction$reduction_func"(i8* [[TMP35]], i8* [[TMP36]]) #[[ATTR4:[0-9]+]]
// CHECK-NEXT: br label [[IFCONT:%.*]]
// CHECK: else:
// CHECK-NEXT: br label [[IFCONT]]
// CHECK: ifcont:
// CHECK-NEXT: [[TMP37:%.*]] = icmp eq i16 [[TMP8]], 1
// CHECK-NEXT: [[TMP38:%.*]] = icmp uge i16 [[TMP6]], [[TMP7]]
// CHECK-NEXT: [[TMP39:%.*]] = and i1 [[TMP37]], [[TMP38]]
// CHECK-NEXT: br i1 [[TMP39]], label [[THEN4:%.*]], label [[ELSE5:%.*]]
// CHECK: then4:
// CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_REMOTE_REDUCE_LIST]], i64 0, i64 0
// CHECK-NEXT: [[TMP41:%.*]] = load i8*, i8** [[TMP40]], align 8
// CHECK-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK-NEXT: [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8
// CHECK-NEXT: [[TMP44:%.*]] = bitcast i8* [[TMP41]] to double*
// CHECK-NEXT: [[TMP45:%.*]] = bitcast i8* [[TMP43]] to double*
// CHECK-NEXT: [[TMP46:%.*]] = load double, double* [[TMP44]], align 8
// CHECK-NEXT: store double [[TMP46]], double* [[TMP45]], align 8
// CHECK-NEXT: br label [[IFCONT6:%.*]]
// CHECK: else5:
// CHECK-NEXT: br label [[IFCONT6]]
// CHECK: ifcont6:
// CHECK-NEXT: ret void
//
//
// CHECK-LABEL: define {{[^@]+}}@_omp_reduction_inter_warp_copy_func
// CHECK-SAME: (i8* [[TMP0:%.*]], i32 [[TMP1:%.*]]) #[[ATTR2]] {
// CHECK-NEXT: entry:
// CHECK-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK-NEXT: [[DOTCNT_ADDR:%.*]] = alloca i32, align 4
// CHECK-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
// CHECK-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK-NEXT: store i32 [[TMP1]], i32* [[DOTADDR1]], align 4
// CHECK-NEXT: [[NVPTX_TID:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block()
// CHECK-NEXT: [[NVPTX_TID2:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block()
// CHECK-NEXT: [[NVPTX_LANE_ID:%.*]] = and i32 [[NVPTX_TID2]], 31
// CHECK-NEXT: [[NVPTX_TID3:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block()
// CHECK-NEXT: [[NVPTX_WARP_ID:%.*]] = ashr i32 [[NVPTX_TID3]], 5
// CHECK-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK-NEXT: [[TMP4:%.*]] = bitcast i8* [[TMP3]] to [1 x i8*]*
// CHECK-NEXT: store i32 0, i32* [[DOTCNT_ADDR]], align 4
// CHECK-NEXT: br label [[PRECOND:%.*]]
// CHECK: precond:
// CHECK-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCNT_ADDR]], align 4
// CHECK-NEXT: [[TMP6:%.*]] = icmp ult i32 [[TMP5]], 2
// CHECK-NEXT: br i1 [[TMP6]], label [[BODY:%.*]], label [[EXIT:%.*]]
// CHECK: body:
// CHECK-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP2]])
// CHECK-NEXT: [[WARP_MASTER:%.*]] = icmp eq i32 [[NVPTX_LANE_ID]], 0
// CHECK-NEXT: br i1 [[WARP_MASTER]], label [[THEN:%.*]], label [[ELSE:%.*]]
// CHECK: then:
// CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP4]], i64 0, i64 0
// CHECK-NEXT: [[TMP8:%.*]] = load i8*, i8** [[TMP7]], align 8
// CHECK-NEXT: [[TMP9:%.*]] = bitcast i8* [[TMP8]] to i32*
// CHECK-NEXT: [[TMP10:%.*]] = getelementptr i32, i32* [[TMP9]], i32 [[TMP5]]
// CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [32 x i32], [32 x i32] addrspace(3)* @__openmp_nvptx_data_transfer_temporary_storage, i64 0, i32 [[NVPTX_WARP_ID]]
// CHECK-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP10]], align 4
// CHECK-NEXT: store volatile i32 [[TMP12]], i32 addrspace(3)* [[TMP11]], align 4
// CHECK-NEXT: br label [[IFCONT:%.*]]
// CHECK: else:
// CHECK-NEXT: br label [[IFCONT]]
// CHECK: ifcont:
// CHECK-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]])
// CHECK-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTADDR1]], align 4
// CHECK-NEXT: [[IS_ACTIVE_THREAD:%.*]] = icmp ult i32 [[NVPTX_TID]], [[TMP13]]
// CHECK-NEXT: br i1 [[IS_ACTIVE_THREAD]], label [[THEN2:%.*]], label [[ELSE3:%.*]]
// CHECK: then2:
// CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds [32 x i32], [32 x i32] addrspace(3)* @__openmp_nvptx_data_transfer_temporary_storage, i64 0, i32 [[NVPTX_TID]]
// CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP4]], i64 0, i64 0
// CHECK-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8
// CHECK-NEXT: [[TMP17:%.*]] = bitcast i8* [[TMP16]] to i32*
// CHECK-NEXT: [[TMP18:%.*]] = getelementptr i32, i32* [[TMP17]], i32 [[TMP5]]
// CHECK-NEXT: [[TMP19:%.*]] = load volatile i32, i32 addrspace(3)* [[TMP14]], align 4
// CHECK-NEXT: store i32 [[TMP19]], i32* [[TMP18]], align 4
// CHECK-NEXT: br label [[IFCONT4:%.*]]
// CHECK: else3:
// CHECK-NEXT: br label [[IFCONT4]]
// CHECK: ifcont4:
// CHECK-NEXT: [[TMP20:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK-NEXT: store i32 [[TMP20]], i32* [[DOTCNT_ADDR]], align 4
// CHECK-NEXT: br label [[PRECOND]]
// CHECK: exit:
// CHECK-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z3barv
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[O:%.*]] = alloca [5 x %class.S2], align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[B:%.*]] = alloca [10 x [10 x [10 x double]]], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [5 x %class.S2], [5 x %class.S2]* [[O]], i32 0, i32 0
// CHECK1-NEXT: [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[CLASS_S2:%.*]], %class.S2* [[ARRAY_BEGIN]], i64 5
// CHECK1-NEXT: br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1: arrayctor.loop:
// CHECK1-NEXT: [[ARRAYCTOR_CUR:%.*]] = phi %class.S2* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT: call void @_ZN2S2C1Ev(%class.S2* nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT: [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[CLASS_S2]], %class.S2* [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT: [[ARRAYCTOR_DONE:%.*]] = icmp eq %class.S2* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT: br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1: arrayctor.cont:
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [5 x %class.S2], [5 x %class.S2]* [[O]], i64 0, i64 0
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP1:%.*]] = bitcast i8** [[TMP0]] to [5 x %class.S2]**
// CHECK1-NEXT: store [5 x %class.S2]* [[O]], [5 x %class.S2]** [[TMP1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to %class.S2**
// CHECK1-NEXT: store %class.S2* [[ARRAYIDX]], %class.S2** [[TMP3]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3barv_l50.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK1-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3barv_l50([5 x %class.S2]* [[O]]) #[[ATTR8:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x [10 x [10 x double]]]**
// CHECK1-NEXT: store [10 x [10 x [10 x double]]]* [[B]], [10 x [10 x [10 x double]]]** [[TMP10]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK1-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x [10 x [10 x double]]]**
// CHECK1-NEXT: store [10 x [10 x [10 x double]]]* [[B]], [10 x [10 x [10 x double]]]** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3barv_l55.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.3, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED4:%.*]], label [[OMP_OFFLOAD_CONT5:%.*]]
// CHECK1: omp_offload.failed4:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3barv_l55([10 x [10 x [10 x double]]]* [[B]]) #[[ATTR8]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT5]]
// CHECK1: omp_offload.cont5:
// CHECK1-NEXT: ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2S2C1Ev
// CHECK1-SAME: (%class.S2* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %class.S2*, align 8
// CHECK1-NEXT: store %class.S2* [[THIS]], %class.S2** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[THIS1:%.*]] = load %class.S2*, %class.S2** [[THIS_ADDR]], align 8
// CHECK1-NEXT: call void @_ZN2S2C2Ev(%class.S2* nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3barv_l50
// CHECK1-SAME: ([5 x %class.S2]* nonnull align 4 dereferenceable(20) [[O:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[O_ADDR:%.*]] = alloca [5 x %class.S2]*, align 8
// CHECK1-NEXT: store [5 x %class.S2]* [[O]], [5 x %class.S2]** [[O_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load [5 x %class.S2]*, [5 x %class.S2]** [[O_ADDR]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [5 x %class.S2]*)* @.omp_outlined. to void (i32*, i32*, ...)*), [5 x %class.S2]* [[TMP0]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [5 x %class.S2]* nonnull align 4 dereferenceable(20) [[O:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[O_ADDR:%.*]] = alloca [5 x %class.S2]*, align 8
// CHECK1-NEXT: [[O1:%.*]] = alloca [[CLASS_S2:%.*]], align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store [5 x %class.S2]* [[O]], [5 x %class.S2]** [[O_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load [5 x %class.S2]*, [5 x %class.S2]** [[O_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [5 x %class.S2], [5 x %class.S2]* [[TMP0]], i64 0, i64 0
// CHECK1-NEXT: call void @_ZN2S2C1Ev(%class.S2* nonnull align 4 dereferenceable(4) [[O1]])
// CHECK1-NEXT: [[TMP1:%.*]] = bitcast [5 x %class.S2]* [[TMP0]] to %class.S2*
// CHECK1-NEXT: [[TMP2:%.*]] = ptrtoint %class.S2* [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = ptrtoint %class.S2* [[ARRAYIDX]] to i64
// CHECK1-NEXT: [[TMP4:%.*]] = sub i64 [[TMP2]], [[TMP3]]
// CHECK1-NEXT: [[TMP5:%.*]] = sdiv exact i64 [[TMP4]], ptrtoint (%class.S2* getelementptr ([[CLASS_S2]], %class.S2* null, i32 1) to i64)
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr [[CLASS_S2]], %class.S2* [[O1]], i64 [[TMP5]]
// CHECK1-NEXT: [[TMP7:%.*]] = bitcast %class.S2* [[TMP6]] to [5 x %class.S2]*
// CHECK1-NEXT: store i32 0, i32* [[I]], align 4
// CHECK1-NEXT: br label [[FOR_COND:%.*]]
// CHECK1: for.cond:
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP8]], 10
// CHECK1-NEXT: br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// CHECK1: for.body:
// CHECK1-NEXT: br label [[FOR_INC:%.*]]
// CHECK1: for.inc:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK1-NEXT: store i32 [[INC]], i32* [[I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast %class.S2* [[O1]] to i8*
// CHECK1-NEXT: store i8* [[TMP11]], i8** [[TMP10]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP13]], i32 1, i64 8, i8* [[TMP14]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(4) %class.S2* @_ZN2S2plERS_(%class.S2* nonnull align 4 dereferenceable(4) [[ARRAYIDX]], %class.S2* nonnull align 4 dereferenceable(4) [[O1]])
// CHECK1-NEXT: [[TMP16:%.*]] = bitcast %class.S2* [[ARRAYIDX]] to i8*
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast %class.S2* [[CALL]] to i8*
// CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP16]], i8* align 4 [[TMP17]], i64 4, i1 false)
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4
// CHECK1-NEXT: call void @__kmpc_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP19]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT: [[CALL2:%.*]] = call nonnull align 4 dereferenceable(4) %class.S2* @_ZN2S2plERS_(%class.S2* nonnull align 4 dereferenceable(4) [[ARRAYIDX]], %class.S2* nonnull align 4 dereferenceable(4) [[O1]])
// CHECK1-NEXT: [[TMP20:%.*]] = bitcast %class.S2* [[ARRAYIDX]] to i8*
// CHECK1-NEXT: [[TMP21:%.*]] = bitcast %class.S2* [[CALL2]] to i8*
// CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP20]], i8* align 4 [[TMP21]], i64 4, i1 false)
// CHECK1-NEXT: call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB2]], i32 [[TMP19]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %class.S2*
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %class.S2*
// CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 4 dereferenceable(4) %class.S2* @_ZN2S2plERS_(%class.S2* nonnull align 4 dereferenceable(4) [[TMP11]], %class.S2* nonnull align 4 dereferenceable(4) [[TMP8]])
// CHECK1-NEXT: [[TMP12:%.*]] = bitcast %class.S2* [[TMP11]] to i8*
// CHECK1-NEXT: [[TMP13:%.*]] = bitcast %class.S2* [[CALL]] to i8*
// CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP12]], i8* align 4 [[TMP13]], i64 4, i1 false)
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3barv_l55
// CHECK1-SAME: ([10 x [10 x [10 x double]]]* nonnull align 8 dereferenceable(8000) [[B:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x [10 x [10 x double]]]*, align 8
// CHECK1-NEXT: store [10 x [10 x [10 x double]]]* [[B]], [10 x [10 x [10 x double]]]** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load [10 x [10 x [10 x double]]]*, [10 x [10 x [10 x double]]]** [[B_ADDR]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x [10 x [10 x double]]]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), [10 x [10 x [10 x double]]]* [[TMP0]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x [10 x [10 x double]]]* nonnull align 8 dereferenceable(8000) [[B:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x [10 x [10 x double]]]*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTRD_INPUT_:%.*]] = alloca [1 x %struct.kmp_taskred_input_t], align 8
// CHECK1-NEXT: [[DOTTASK_RED_:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[I:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca double, align 8
// CHECK1-NEXT: [[_TMP30:%.*]] = alloca double, align 8
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store [10 x [10 x [10 x double]]]* [[B]], [10 x [10 x [10 x double]]]** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load [10 x [10 x [10 x double]]]*, [10 x [10 x [10 x double]]]** [[B_ADDR]], align 8
// CHECK1-NEXT: store i64 0, i64* [[DOTOMP_LB]], align 8
// CHECK1-NEXT: store i64 9, i64* [[DOTOMP_UB]], align 8
// CHECK1-NEXT: store i64 1, i64* [[DOTOMP_STRIDE]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [10 x [10 x double]]], [10 x [10 x [10 x double]]]* [[TMP0]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [10 x [10 x double]], [10 x [10 x double]]* [[ARRAYIDX]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds [10 x double], [10 x double]* [[ARRAYDECAY]], i64 2
// CHECK1-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [10 x double], [10 x double]* [[ARRAYIDX1]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds double, double* [[ARRAYDECAY2]], i64 1
// CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [10 x [10 x [10 x double]]], [10 x [10 x [10 x double]]]* [[TMP0]], i64 0, i64 1
// CHECK1-NEXT: [[ARRAYDECAY5:%.*]] = getelementptr inbounds [10 x [10 x double]], [10 x [10 x double]]* [[ARRAYIDX4]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [10 x double], [10 x double]* [[ARRAYDECAY5]], i64 5
// CHECK1-NEXT: [[ARRAYDECAY7:%.*]] = getelementptr inbounds [10 x double], [10 x double]* [[ARRAYIDX6]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds double, double* [[ARRAYDECAY7]], i64 1
// CHECK1-NEXT: [[TMP1:%.*]] = ptrtoint double* [[ARRAYIDX8]] to i64
// CHECK1-NEXT: [[TMP2:%.*]] = ptrtoint double* [[ARRAYIDX3]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = sub i64 [[TMP1]], [[TMP2]]
// CHECK1-NEXT: [[TMP4:%.*]] = sdiv exact i64 [[TMP3]], ptrtoint (double* getelementptr (double, double* null, i32 1) to i64)
// CHECK1-NEXT: [[TMP5:%.*]] = add nuw i64 [[TMP4]], 1
// CHECK1-NEXT: [[TMP6:%.*]] = mul nuw i64 [[TMP5]], ptrtoint (double* getelementptr (double, double* null, i32 1) to i64)
// CHECK1-NEXT: [[TMP7:%.*]] = call i8* @llvm.stacksave()
// CHECK1-NEXT: store i8* [[TMP7]], i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: [[VLA:%.*]] = alloca double, i64 [[TMP5]], align 8
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[__VLA_EXPR0]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr double, double* [[VLA]], i64 [[TMP5]]
// CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq double* [[VLA]], [[TMP8]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1: omp.arrayinit.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi double* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT: store double 0.000000e+00, double* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 8
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr double, double* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq double* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1: omp.arrayinit.done:
// CHECK1-NEXT: [[TMP9:%.*]] = bitcast [10 x [10 x [10 x double]]]* [[TMP0]] to double*
// CHECK1-NEXT: [[TMP10:%.*]] = ptrtoint double* [[TMP9]] to i64
// CHECK1-NEXT: [[TMP11:%.*]] = ptrtoint double* [[ARRAYIDX3]] to i64
// CHECK1-NEXT: [[TMP12:%.*]] = sub i64 [[TMP10]], [[TMP11]]
// CHECK1-NEXT: [[TMP13:%.*]] = sdiv exact i64 [[TMP12]], ptrtoint (double* getelementptr (double, double* null, i32 1) to i64)
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr double, double* [[VLA]], i64 [[TMP13]]
// CHECK1-NEXT: [[TMP15:%.*]] = bitcast double* [[TMP14]] to [10 x [10 x [10 x double]]]*
// CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [1 x %struct.kmp_taskred_input_t], [1 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 0
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0
// CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [10 x [10 x [10 x double]]], [10 x [10 x [10 x double]]]* [[TMP0]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYDECAY10:%.*]] = getelementptr inbounds [10 x [10 x double]], [10 x [10 x double]]* [[ARRAYIDX9]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [10 x double], [10 x double]* [[ARRAYDECAY10]], i64 2
// CHECK1-NEXT: [[ARRAYDECAY12:%.*]] = getelementptr inbounds [10 x double], [10 x double]* [[ARRAYIDX11]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds double, double* [[ARRAYDECAY12]], i64 1
// CHECK1-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [10 x [10 x [10 x double]]], [10 x [10 x [10 x double]]]* [[TMP0]], i64 0, i64 1
// CHECK1-NEXT: [[ARRAYDECAY15:%.*]] = getelementptr inbounds [10 x [10 x double]], [10 x [10 x double]]* [[ARRAYIDX14]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds [10 x double], [10 x double]* [[ARRAYDECAY15]], i64 5
// CHECK1-NEXT: [[ARRAYDECAY17:%.*]] = getelementptr inbounds [10 x double], [10 x double]* [[ARRAYIDX16]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX18:%.*]] = getelementptr inbounds double, double* [[ARRAYDECAY17]], i64 1
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast double* [[VLA]] to i8*
// CHECK1-NEXT: store i8* [[TMP17]], i8** [[TMP16]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1
// CHECK1-NEXT: [[TMP19:%.*]] = bitcast double* [[ARRAYIDX13]] to i8*
// CHECK1-NEXT: store i8* [[TMP19]], i8** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = ptrtoint double* [[ARRAYIDX18]] to i64
// CHECK1-NEXT: [[TMP21:%.*]] = ptrtoint double* [[ARRAYIDX13]] to i64
// CHECK1-NEXT: [[TMP22:%.*]] = sub i64 [[TMP20]], [[TMP21]]
// CHECK1-NEXT: [[TMP23:%.*]] = sdiv exact i64 [[TMP22]], ptrtoint (double* getelementptr (double, double* null, i32 1) to i64)
// CHECK1-NEXT: [[TMP24:%.*]] = add nuw i64 [[TMP23]], 1
// CHECK1-NEXT: [[TMP25:%.*]] = mul nuw i64 [[TMP24]], ptrtoint (double* getelementptr (double, double* null, i32 1) to i64)
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2
// CHECK1-NEXT: store i64 [[TMP25]], i64* [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3
// CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP27]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4
// CHECK1-NEXT: store i8* null, i8** [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5
// CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6
// CHECK1-NEXT: store i32 1, i32* [[TMP30]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4
// CHECK1-NEXT: [[TMP33:%.*]] = bitcast [1 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8*
// CHECK1-NEXT: [[TMP34:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB2]], i32 [[TMP32]], i32 1, i32 1, i8* [[TMP33]])
// CHECK1-NEXT: store i8* [[TMP34]], i8** [[DOTTASK_RED_]], align 8
// CHECK1-NEXT: [[TMP35:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP36:%.*]] = load i32, i32* [[TMP35]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_8(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP36]], i32 34, i32* [[DOTOMP_IS_LAST]], i64* [[DOTOMP_LB]], i64* [[DOTOMP_UB]], i64* [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK1-NEXT: [[TMP37:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP37]], 9
// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1: cond.true:
// CHECK1-NEXT: br label [[COND_END:%.*]]
// CHECK1: cond.false:
// CHECK1-NEXT: [[TMP38:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP38]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK1-NEXT: store i64 [[TMP39]], i64* [[DOTOMP_IV]], align 8
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP40:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[TMP41:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[CMP19:%.*]] = icmp sle i64 [[TMP40]], [[TMP41]]
// CHECK1-NEXT: br i1 [[CMP19]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1: omp.inner.for.cond.cleanup:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP42]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL]]
// CHECK1-NEXT: store i64 [[ADD]], i64* [[I]], align 8
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[ADD20:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK1-NEXT: store i64 [[ADD20]], i64* [[DOTOMP_IV]], align 8
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
// CHECK1-NEXT: [[TMP44:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP45:%.*]] = load i32, i32* [[TMP44]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB3]], i32 [[TMP45]])
// CHECK1-NEXT: [[TMP46:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP47:%.*]] = load i32, i32* [[TMP46]], align 4
// CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP47]], i32 1)
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: [[TMP49:%.*]] = bitcast double* [[VLA]] to i8*
// CHECK1-NEXT: store i8* [[TMP49]], i8** [[TMP48]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT: [[TMP51:%.*]] = inttoptr i64 [[TMP5]] to i8*
// CHECK1-NEXT: store i8* [[TMP51]], i8** [[TMP50]], align 8
// CHECK1-NEXT: [[TMP52:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP53:%.*]] = load i32, i32* [[TMP52]], align 4
// CHECK1-NEXT: [[TMP54:%.*]] = bitcast [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT: [[TMP55:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP53]], i32 1, i64 16, i8* [[TMP54]], void (i8*, i8*)* @.omp.reduction.reduction_func.2, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP55]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP56:%.*]] = getelementptr double, double* [[ARRAYIDX3]], i64 [[TMP5]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq double* [[ARRAYIDX3]], [[TMP56]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE25:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi double* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST21:%.*]] = phi double* [ [[ARRAYIDX3]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT23:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP57:%.*]] = load double, double* [[OMP_ARRAYCPY_DESTELEMENTPAST21]], align 8
// CHECK1-NEXT: [[TMP58:%.*]] = load double, double* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 8
// CHECK1-NEXT: [[ADD22:%.*]] = fadd double [[TMP57]], [[TMP58]]
// CHECK1-NEXT: store double [[ADD22]], double* [[OMP_ARRAYCPY_DESTELEMENTPAST21]], align 8
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT23]] = getelementptr double, double* [[OMP_ARRAYCPY_DESTELEMENTPAST21]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr double, double* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE24:%.*]] = icmp eq double* [[OMP_ARRAYCPY_DEST_ELEMENT23]], [[TMP56]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE24]], label [[OMP_ARRAYCPY_DONE25]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done25:
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB1]], i32 [[TMP53]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP59:%.*]] = getelementptr double, double* [[ARRAYIDX3]], i64 [[TMP5]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY26:%.*]] = icmp eq double* [[ARRAYIDX3]], [[TMP59]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY26]], label [[OMP_ARRAYCPY_DONE35:%.*]], label [[OMP_ARRAYCPY_BODY27:%.*]]
// CHECK1: omp.arraycpy.body27:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST28:%.*]] = phi double* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT33:%.*]], [[ATOMIC_EXIT:%.*]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST29:%.*]] = phi double* [ [[ARRAYIDX3]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT32:%.*]], [[ATOMIC_EXIT]] ]
// CHECK1-NEXT: [[TMP60:%.*]] = load double, double* [[OMP_ARRAYCPY_SRCELEMENTPAST28]], align 8
// CHECK1-NEXT: [[TMP61:%.*]] = bitcast double* [[OMP_ARRAYCPY_DESTELEMENTPAST29]] to i64*
// CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i64, i64* [[TMP61]] monotonic, align 8
// CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK1: atomic_cont:
// CHECK1-NEXT: [[TMP62:%.*]] = phi i64 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY27]] ], [ [[TMP70:%.*]], [[ATOMIC_CONT]] ]
// CHECK1-NEXT: [[TMP63:%.*]] = bitcast double* [[ATOMIC_TEMP]] to i64*
// CHECK1-NEXT: [[TMP64:%.*]] = bitcast i64 [[TMP62]] to double
// CHECK1-NEXT: store double [[TMP64]], double* [[_TMP30]], align 8
// CHECK1-NEXT: [[TMP65:%.*]] = load double, double* [[_TMP30]], align 8
// CHECK1-NEXT: [[TMP66:%.*]] = load double, double* [[OMP_ARRAYCPY_SRCELEMENTPAST28]], align 8
// CHECK1-NEXT: [[ADD31:%.*]] = fadd double [[TMP65]], [[TMP66]]
// CHECK1-NEXT: store double [[ADD31]], double* [[ATOMIC_TEMP]], align 8
// CHECK1-NEXT: [[TMP67:%.*]] = load i64, i64* [[TMP63]], align 8
// CHECK1-NEXT: [[TMP68:%.*]] = bitcast double* [[OMP_ARRAYCPY_DESTELEMENTPAST29]] to i64*
// CHECK1-NEXT: [[TMP69:%.*]] = cmpxchg i64* [[TMP68]], i64 [[TMP62]], i64 [[TMP67]] monotonic monotonic, align 8
// CHECK1-NEXT: [[TMP70]] = extractvalue { i64, i1 } [[TMP69]], 0
// CHECK1-NEXT: [[TMP71:%.*]] = extractvalue { i64, i1 } [[TMP69]], 1
// CHECK1-NEXT: br i1 [[TMP71]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]]
// CHECK1: atomic_exit:
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT32]] = getelementptr double, double* [[OMP_ARRAYCPY_DESTELEMENTPAST29]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT33]] = getelementptr double, double* [[OMP_ARRAYCPY_SRCELEMENTPAST28]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE34:%.*]] = icmp eq double* [[OMP_ARRAYCPY_DEST_ELEMENT32]], [[TMP59]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE34]], label [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_BODY27]]
// CHECK1: omp.arraycpy.done35:
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default:
// CHECK1-NEXT: [[TMP72:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP72]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.red_init.
// CHECK1-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8
// CHECK1-NEXT: [[TMP4:%.*]] = bitcast i8* [[TMP2]] to double*
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr double, double* [[TMP4]], i64 [[TMP3]]
// CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq double* [[TMP4]], [[TMP5]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1: omp.arrayinit.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi double* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT: store double 0.000000e+00, double* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 8
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr double, double* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq double* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1: omp.arrayinit.done:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.red_comb.
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = bitcast i8* [[TMP3]] to double*
// CHECK1-NEXT: [[TMP5:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = bitcast i8* [[TMP5]] to double*
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr double, double* [[TMP4]], i64 [[TMP2]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq double* [[TMP4]], [[TMP7]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi double* [ [[TMP6]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi double* [ [[TMP4]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP8:%.*]] = load double, double* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = load double, double* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 8
// CHECK1-NEXT: [[ADD:%.*]] = fadd double [[TMP8]], [[TMP9]]
// CHECK1-NEXT: store double [[ADD]], double* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 8
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr double, double* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr double, double* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq double* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP7]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done2:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [2 x i8*]*
// CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [2 x i8*]*
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to double*
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to double*
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr double, double* [[TMP11]], i64 [[TMP14]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq double* [[TMP11]], [[TMP15]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi double* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi double* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP16:%.*]] = load double, double* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = load double, double* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 8
// CHECK1-NEXT: [[ADD:%.*]] = fadd double [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store double [[ADD]], double* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 8
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr double, double* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr double, double* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq double* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP15]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done2:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2S2C2Ev
// CHECK1-SAME: (%class.S2* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %class.S2*, align 8
// CHECK1-NEXT: store %class.S2* [[THIS]], %class.S2** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[THIS1:%.*]] = load %class.S2*, %class.S2** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[CLASS_S2:%.*]], %class.S2* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: store i32 0, i32* [[A]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: () #[[ATTR10:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT: [[CALL:%.*]] = call signext i32 @_Z3barv()
// CHECK1-NEXT: store i32 [[CALL]], i32* [[A]], align 4
// CHECK1-NEXT: ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR11:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_Z3sumPiiS_
// CHECK2-SAME: (i32* [[INPUT:%.*]], i32 [[SIZE:%.*]], i32* [[OUTPUT:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[INPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[OUTPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[SIZE_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SIZE_CASTED4:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS7:%.*]] = alloca [3 x i8*], align 4
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS8:%.*]] = alloca [3 x i8*], align 4
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS9:%.*]] = alloca [3 x i8*], align 4
// CHECK2-NEXT: [[_TMP10:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_12:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A:%.*]] = alloca [10 x i32], align 4
// CHECK2-NEXT: [[SIZE_CASTED19:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [2 x i8*], align 4
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [2 x i8*], align 4
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [2 x i8*], align 4
// CHECK2-NEXT: [[SIZE_CASTED26:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS28:%.*]] = alloca [2 x i8*], align 4
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS29:%.*]] = alloca [2 x i8*], align 4
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS30:%.*]] = alloca [2 x i8*], align 4
// CHECK2-NEXT: store i32* [[INPUT]], i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32* [[OUTPUT]], i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP0]], i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP5]], i32 0
// CHECK2-NEXT: [[TMP6:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, i32* [[TMP7]], i32 0
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK2-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[TMP12]], align 4
// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[TMP14]], align 4
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK2-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK2-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK2-NEXT: store i32* [[TMP4]], i32** [[TMP17]], align 4
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK2-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK2-NEXT: store i32* [[ARRAYIDX]], i32** [[TMP19]], align 4
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32**
// CHECK2-NEXT: store i32* [[TMP6]], i32** [[TMP22]], align 4
// CHECK2-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32**
// CHECK2-NEXT: store i32* [[ARRAYIDX1]], i32** [[TMP24]], align 4
// CHECK2-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK2-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP28]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP29]], 0
// CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK2-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK2-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP30]], 1
// CHECK2-NEXT: [[TMP31:%.*]] = zext i32 [[ADD]] to i64
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP31]])
// CHECK2-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l69.region_id, i32 3, i8** [[TMP26]], i8** [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// CHECK2-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l69(i32 [[TMP1]], i32* [[TMP2]], i32* [[TMP3]]) #[[ATTR2:[0-9]+]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont:
// CHECK2-NEXT: [[TMP34:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP34]], i32* [[SIZE_CASTED4]], align 4
// CHECK2-NEXT: [[TMP35:%.*]] = load i32, i32* [[SIZE_CASTED4]], align 4
// CHECK2-NEXT: [[TMP36:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP37:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP38:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP39:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[TMP39]], i32 0
// CHECK2-NEXT: [[TMP40:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP41:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, i32* [[TMP41]], i32 0
// CHECK2-NEXT: [[TMP42:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: [[TMP43:%.*]] = mul nuw i32 [[TMP42]], 4
// CHECK2-NEXT: [[TMP44:%.*]] = sext i32 [[TMP43]] to i64
// CHECK2-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK2-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32*
// CHECK2-NEXT: store i32 [[TMP35]], i32* [[TMP46]], align 4
// CHECK2-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK2-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32*
// CHECK2-NEXT: store i32 [[TMP35]], i32* [[TMP48]], align 4
// CHECK2-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0
// CHECK2-NEXT: store i8* null, i8** [[TMP49]], align 4
// CHECK2-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 1
// CHECK2-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32**
// CHECK2-NEXT: store i32* [[TMP38]], i32** [[TMP51]], align 4
// CHECK2-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK2-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to i32**
// CHECK2-NEXT: store i32* [[ARRAYIDX5]], i32** [[TMP53]], align 4
// CHECK2-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP54]], align 4
// CHECK2-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 2
// CHECK2-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to i32**
// CHECK2-NEXT: store i32* [[TMP40]], i32** [[TMP56]], align 4
// CHECK2-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 2
// CHECK2-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to i32**
// CHECK2-NEXT: store i32* [[ARRAYIDX6]], i32** [[TMP58]], align 4
// CHECK2-NEXT: store i64 [[TMP44]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.7, i32 0, i32 2), align 4
// CHECK2-NEXT: [[TMP59:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 2
// CHECK2-NEXT: store i8* null, i8** [[TMP59]], align 4
// CHECK2-NEXT: [[TMP60:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK2-NEXT: [[TMP61:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK2-NEXT: [[TMP62:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP62]], i32* [[DOTCAPTURE_EXPR_11]], align 4
// CHECK2-NEXT: [[TMP63:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4
// CHECK2-NEXT: [[SUB13:%.*]] = sub nsw i32 [[TMP63]], 0
// CHECK2-NEXT: [[DIV14:%.*]] = sdiv i32 [[SUB13]], 1
// CHECK2-NEXT: [[SUB15:%.*]] = sub nsw i32 [[DIV14]], 1
// CHECK2-NEXT: store i32 [[SUB15]], i32* [[DOTCAPTURE_EXPR_12]], align 4
// CHECK2-NEXT: [[TMP64:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_12]], align 4
// CHECK2-NEXT: [[ADD16:%.*]] = add nsw i32 [[TMP64]], 1
// CHECK2-NEXT: [[TMP65:%.*]] = zext i32 [[ADD16]] to i64
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i64 [[TMP65]])
// CHECK2-NEXT: [[TMP66:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l73.region_id, i32 3, i8** [[TMP60]], i8** [[TMP61]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.7, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP67:%.*]] = icmp ne i32 [[TMP66]], 0
// CHECK2-NEXT: br i1 [[TMP67]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK2: omp_offload.failed17:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l73(i32 [[TMP35]], i32* [[TMP36]], i32* [[TMP37]]) #[[ATTR2]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK2: omp_offload.cont18:
// CHECK2-NEXT: [[TMP68:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP68]], i32* [[SIZE_CASTED19]], align 4
// CHECK2-NEXT: [[TMP69:%.*]] = load i32, i32* [[SIZE_CASTED19]], align 4
// CHECK2-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[A]], i32 0, i32 0
// CHECK2-NEXT: [[TMP70:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK2-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK2-NEXT: store i32 [[TMP69]], i32* [[TMP71]], align 4
// CHECK2-NEXT: [[TMP72:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK2-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32*
// CHECK2-NEXT: store i32 [[TMP69]], i32* [[TMP73]], align 4
// CHECK2-NEXT: [[TMP74:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 0
// CHECK2-NEXT: store i8* null, i8** [[TMP74]], align 4
// CHECK2-NEXT: [[TMP75:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK2-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [10 x i32]**
// CHECK2-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP76]], align 4
// CHECK2-NEXT: [[TMP77:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK2-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32**
// CHECK2-NEXT: store i32* [[ARRAYIDX20]], i32** [[TMP78]], align 4
// CHECK2-NEXT: [[TMP79:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP79]], align 4
// CHECK2-NEXT: [[TMP80:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK2-NEXT: [[TMP81:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK2-NEXT: [[TMP82:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l78.region_id, i32 2, i8** [[TMP80]], i8** [[TMP81]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP83:%.*]] = icmp ne i32 [[TMP82]], 0
// CHECK2-NEXT: br i1 [[TMP83]], label [[OMP_OFFLOAD_FAILED24:%.*]], label [[OMP_OFFLOAD_CONT25:%.*]]
// CHECK2: omp_offload.failed24:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l78(i32 [[TMP69]], [10 x i32]* [[A]]) #[[ATTR2]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT25]]
// CHECK2: omp_offload.cont25:
// CHECK2-NEXT: [[TMP84:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP84]], i32* [[SIZE_CASTED26]], align 4
// CHECK2-NEXT: [[TMP85:%.*]] = load i32, i32* [[SIZE_CASTED26]], align 4
// CHECK2-NEXT: [[ARRAYIDX27:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[A]], i32 0, i32 3
// CHECK2-NEXT: [[TMP86:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS28]], i32 0, i32 0
// CHECK2-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to i32*
// CHECK2-NEXT: store i32 [[TMP85]], i32* [[TMP87]], align 4
// CHECK2-NEXT: [[TMP88:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS29]], i32 0, i32 0
// CHECK2-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to i32*
// CHECK2-NEXT: store i32 [[TMP85]], i32* [[TMP89]], align 4
// CHECK2-NEXT: [[TMP90:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS30]], i32 0, i32 0
// CHECK2-NEXT: store i8* null, i8** [[TMP90]], align 4
// CHECK2-NEXT: [[TMP91:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS28]], i32 0, i32 1
// CHECK2-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to [10 x i32]**
// CHECK2-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP92]], align 4
// CHECK2-NEXT: [[TMP93:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS29]], i32 0, i32 1
// CHECK2-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i32**
// CHECK2-NEXT: store i32* [[ARRAYIDX27]], i32** [[TMP94]], align 4
// CHECK2-NEXT: [[TMP95:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS30]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP95]], align 4
// CHECK2-NEXT: [[TMP96:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS28]], i32 0, i32 0
// CHECK2-NEXT: [[TMP97:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS29]], i32 0, i32 0
// CHECK2-NEXT: [[TMP98:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l81.region_id, i32 2, i8** [[TMP96]], i8** [[TMP97]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP99:%.*]] = icmp ne i32 [[TMP98]], 0
// CHECK2-NEXT: br i1 [[TMP99]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK2: omp_offload.failed31:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l81(i32 [[TMP85]], [10 x i32]* [[A]]) #[[ATTR2]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK2: omp_offload.cont32:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l69
// CHECK2-SAME: (i32 [[SIZE:%.*]], i32* [[OUTPUT:%.*]], i32* [[INPUT:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[OUTPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[INPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[SIZE_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32* [[OUTPUT]], i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: store i32* [[INPUT]], i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP0]], i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB4]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32 [[TMP1]], i32* [[TMP2]], i32* [[TMP3]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[SIZE:%.*]], i32* [[OUTPUT:%.*]], i32* [[INPUT:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[OUTPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[INPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[OUTPUT1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I5:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SIZE_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32* [[OUTPUT]], i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: store i32* [[INPUT]], i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP0]], i32 0
// CHECK2-NEXT: store i32 0, i32* [[OUTPUT1]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = ptrtoint i32* [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = ptrtoint i32* [[ARRAYIDX]] to i64
// CHECK2-NEXT: [[TMP4:%.*]] = sub i64 [[TMP2]], [[TMP3]]
// CHECK2-NEXT: [[TMP5:%.*]] = sdiv exact i64 [[TMP4]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr i32, i32* [[OUTPUT1]], i64 [[TMP5]]
// CHECK2-NEXT: store i32* [[TMP6]], i32** [[TMP]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP7]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP8]], 0
// CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK2-NEXT: [[SUB4:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK2-NEXT: store i32 [[SUB4]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK2-NEXT: store i32 0, i32* [[I]], align 4
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP9]]
// CHECK2-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK2: omp.precond.then:
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK2-NEXT: store i32 [[TMP10]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP12]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK2-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP13]], [[TMP14]]
// CHECK2-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2: cond.true:
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK2-NEXT: br label [[COND_END:%.*]]
// CHECK2: cond.false:
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP15]], [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
// CHECK2-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP22]], i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP23:%.*]] = load i32, i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP24:%.*]] = load i32*, i32** [[TMP]], align 4
// CHECK2-NEXT: [[TMP25:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB4]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP20]], i32 [[TMP21]], i32 [[TMP23]], i32* [[TMP24]], i32* [[TMP25]])
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP26]], [[TMP27]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
// CHECK2-NEXT: [[TMP28:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP29:%.*]] = load i32, i32* [[TMP28]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP29]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK2-NEXT: [[TMP31:%.*]] = bitcast i32* [[OUTPUT1]] to i8*
// CHECK2-NEXT: store i8* [[TMP31]], i8** [[TMP30]], align 4
// CHECK2-NEXT: [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4
// CHECK2-NEXT: [[TMP34:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT: [[TMP35:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP33]], i32 1, i32 4, i8* [[TMP34]], void (i8*, i8*)* @.omp.reduction.reduction_func.2, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: switch i32 [[TMP35]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT: ]
// CHECK2: .omp.reduction.case1:
// CHECK2-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK2-NEXT: [[TMP37:%.*]] = load i32, i32* [[OUTPUT1]], align 4
// CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP36]], [[TMP37]]
// CHECK2-NEXT: store i32 [[ADD8]], i32* [[ARRAYIDX]], align 4
// CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP33]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.case2:
// CHECK2-NEXT: [[TMP38:%.*]] = load i32, i32* [[OUTPUT1]], align 4
// CHECK2-NEXT: [[TMP39:%.*]] = atomicrmw add i32* [[ARRAYIDX]], i32 [[TMP38]] monotonic, align 4
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.default:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32 [[SIZE:%.*]], i32* [[OUTPUT:%.*]], i32* [[INPUT:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[OUTPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[INPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[OUTPUT3:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[_TMP4:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[I5:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK2-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK2-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32* [[OUTPUT]], i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: store i32* [[INPUT]], i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP0]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP1]], 0
// CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK2-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK2-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: store i32 0, i32* [[I]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP2]]
// CHECK2-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK2: omp.precond.then:
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP6]], i32 0
// CHECK2-NEXT: store i32 0, i32* [[OUTPUT3]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = ptrtoint i32* [[TMP7]] to i64
// CHECK2-NEXT: [[TMP9:%.*]] = ptrtoint i32* [[ARRAYIDX]] to i64
// CHECK2-NEXT: [[TMP10:%.*]] = sub i64 [[TMP8]], [[TMP9]]
// CHECK2-NEXT: [[TMP11:%.*]] = sdiv exact i64 [[TMP10]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr i32, i32* [[OUTPUT3]], i64 [[TMP11]]
// CHECK2-NEXT: store i32* [[TMP12]], i32** [[_TMP4]], align 4
// CHECK2-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[TMP13]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP14]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP15]], [[TMP16]]
// CHECK2-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2: cond.true:
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: br label [[COND_END:%.*]]
// CHECK2: cond.false:
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP17]], [[COND_TRUE]] ], [ [[TMP18]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP19]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]]
// CHECK2-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP22]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[I5]], align 4
// CHECK2-NEXT: [[TMP23:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP24:%.*]] = load i32, i32* [[I5]], align 4
// CHECK2-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[TMP23]], i32 [[TMP24]]
// CHECK2-NEXT: [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX8]], align 4
// CHECK2-NEXT: [[TMP26:%.*]] = load i32*, i32** [[_TMP4]], align 4
// CHECK2-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i32 0
// CHECK2-NEXT: [[TMP27:%.*]] = load i32, i32* [[ARRAYIDX9]], align 4
// CHECK2-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP27]], [[TMP25]]
// CHECK2-NEXT: store i32 [[ADD10]], i32* [[ARRAYIDX9]], align 4
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP28]], 1
// CHECK2-NEXT: store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
// CHECK2-NEXT: [[TMP29:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, i32* [[TMP29]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP30]])
// CHECK2-NEXT: [[TMP31:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK2-NEXT: [[TMP32:%.*]] = bitcast i32* [[OUTPUT3]] to i8*
// CHECK2-NEXT: store i8* [[TMP32]], i8** [[TMP31]], align 4
// CHECK2-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4
// CHECK2-NEXT: [[TMP35:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT: [[TMP36:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP34]], i32 1, i32 4, i8* [[TMP35]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: switch i32 [[TMP36]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT: ]
// CHECK2: .omp.reduction.case1:
// CHECK2-NEXT: [[TMP37:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK2-NEXT: [[TMP38:%.*]] = load i32, i32* [[OUTPUT3]], align 4
// CHECK2-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP37]], [[TMP38]]
// CHECK2-NEXT: store i32 [[ADD12]], i32* [[ARRAYIDX]], align 4
// CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP34]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.case2:
// CHECK2-NEXT: [[TMP39:%.*]] = load i32, i32* [[OUTPUT3]], align 4
// CHECK2-NEXT: [[TMP40:%.*]] = atomicrmw add i32* [[ARRAYIDX]], i32 [[TMP39]] monotonic, align 4
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.default:
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 4
// CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i32 0, i32 0
// CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 4
// CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i32 0, i32 0
// CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l73
// CHECK2-SAME: (i32 [[SIZE:%.*]], i32* [[OUTPUT:%.*]], i32* [[INPUT:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[OUTPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[INPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[SIZE_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32* [[OUTPUT]], i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: store i32* [[INPUT]], i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP0]], i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB4]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP1]], i32* [[TMP2]], i32* [[TMP3]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[SIZE:%.*]], i32* [[OUTPUT:%.*]], i32* [[INPUT:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[OUTPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[INPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[OUTPUT2:%.*]] = alloca [3 x i32], align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[_TMP3:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I6:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SIZE_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32* [[OUTPUT]], i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: store i32* [[INPUT]], i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP0]], i32 0
// CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, i32* [[TMP1]], i32 2
// CHECK2-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [3 x i32], [3 x i32]* [[OUTPUT2]], i32 0, i32 0
// CHECK2-NEXT: [[TMP2:%.*]] = getelementptr i32, i32* [[ARRAY_BEGIN]], i32 3
// CHECK2-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i32* [[ARRAY_BEGIN]], [[TMP2]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2: omp.arrayinit.body:
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT: store i32 0, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2: omp.arrayinit.done:
// CHECK2-NEXT: [[TMP3:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = ptrtoint i32* [[TMP3]] to i64
// CHECK2-NEXT: [[TMP5:%.*]] = ptrtoint i32* [[ARRAYIDX]] to i64
// CHECK2-NEXT: [[TMP6:%.*]] = sub i64 [[TMP4]], [[TMP5]]
// CHECK2-NEXT: [[TMP7:%.*]] = sdiv exact i64 [[TMP6]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK2-NEXT: [[TMP8:%.*]] = bitcast [3 x i32]* [[OUTPUT2]] to i32*
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr i32, i32* [[TMP8]], i64 [[TMP7]]
// CHECK2-NEXT: store i32* [[TMP9]], i32** [[TMP]], align 4
// CHECK2-NEXT: [[RHS_BEGIN:%.*]] = bitcast [3 x i32]* [[OUTPUT2]] to i32*
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], 0
// CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK2-NEXT: [[SUB5:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK2-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_4]], align 4
// CHECK2-NEXT: store i32 0, i32* [[I]], align 4
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP12]]
// CHECK2-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK2: omp.precond.then:
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4
// CHECK2-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP14:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP15]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4
// CHECK2-NEXT: [[CMP7:%.*]] = icmp sgt i32 [[TMP16]], [[TMP17]]
// CHECK2-NEXT: br i1 [[CMP7]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2: cond.true:
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4
// CHECK2-NEXT: br label [[COND_END:%.*]]
// CHECK2: cond.false:
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP18]], [[COND_TRUE]] ], [ [[TMP19]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP20]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP21]], [[TMP22]]
// CHECK2-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP25:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP25]], i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP26:%.*]] = load i32, i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP]], align 4
// CHECK2-NEXT: [[TMP28:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB4]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32 [[TMP23]], i32 [[TMP24]], i32 [[TMP26]], i32* [[TMP27]], i32* [[TMP28]])
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
// CHECK2-NEXT: [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP32]])
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: [[TMP33:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK2-NEXT: [[TMP34:%.*]] = bitcast i32* [[RHS_BEGIN]] to i8*
// CHECK2-NEXT: store i8* [[TMP34]], i8** [[TMP33]], align 4
// CHECK2-NEXT: [[TMP35:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP36:%.*]] = load i32, i32* [[TMP35]], align 4
// CHECK2-NEXT: [[TMP37:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT: [[TMP38:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP36]], i32 1, i32 4, i8* [[TMP37]], void (i8*, i8*)* @.omp.reduction.reduction_func.6, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: switch i32 [[TMP38]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT: ]
// CHECK2: .omp.reduction.case1:
// CHECK2-NEXT: [[TMP39:%.*]] = getelementptr i32, i32* [[ARRAYIDX]], i32 3
// CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[ARRAYIDX]], [[TMP39]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE13:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2: omp.arraycpy.body:
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST9:%.*]] = phi i32* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[TMP40:%.*]] = load i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST9]], align 4
// CHECK2-NEXT: [[TMP41:%.*]] = load i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK2-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP40]], [[TMP41]]
// CHECK2-NEXT: store i32 [[ADD10]], i32* [[OMP_ARRAYCPY_DESTELEMENTPAST9]], align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST9]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE12:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP39]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE12]], label [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2: omp.arraycpy.done13:
// CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP36]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.case2:
// CHECK2-NEXT: [[TMP42:%.*]] = getelementptr i32, i32* [[ARRAYIDX]], i32 3
// CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY14:%.*]] = icmp eq i32* [[ARRAYIDX]], [[TMP42]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY14]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY15:%.*]]
// CHECK2: omp.arraycpy.body15:
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST16:%.*]] = phi i32* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT19:%.*]], [[OMP_ARRAYCPY_BODY15]] ]
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST17:%.*]] = phi i32* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT18:%.*]], [[OMP_ARRAYCPY_BODY15]] ]
// CHECK2-NEXT: [[TMP43:%.*]] = load i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST16]], align 4
// CHECK2-NEXT: [[TMP44:%.*]] = atomicrmw add i32* [[OMP_ARRAYCPY_DESTELEMENTPAST17]], i32 [[TMP43]] monotonic, align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT18]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST17]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT19]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST16]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT18]], [[TMP42]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY15]]
// CHECK2: omp.arraycpy.done21:
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.default:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32 [[SIZE:%.*]], i32* [[OUTPUT:%.*]], i32* [[INPUT:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[OUTPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[INPUT_ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[OUTPUT4:%.*]] = alloca [3 x i32], align 4
// CHECK2-NEXT: [[_TMP5:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[I6:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK2-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK2-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32* [[OUTPUT]], i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: store i32* [[INPUT]], i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP0]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP1]], 0
// CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK2-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK2-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: store i32 0, i32* [[I]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP2]]
// CHECK2-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK2: omp.precond.then:
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP6]], i32 0
// CHECK2-NEXT: [[TMP7:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i32, i32* [[TMP7]], i32 2
// CHECK2-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [3 x i32], [3 x i32]* [[OUTPUT4]], i32 0, i32 0
// CHECK2-NEXT: [[TMP8:%.*]] = getelementptr i32, i32* [[ARRAY_BEGIN]], i32 3
// CHECK2-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i32* [[ARRAY_BEGIN]], [[TMP8]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2: omp.arrayinit.body:
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[ARRAY_BEGIN]], [[OMP_PRECOND_THEN]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT: store i32 0, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2: omp.arrayinit.done:
// CHECK2-NEXT: [[TMP9:%.*]] = load i32*, i32** [[OUTPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP10:%.*]] = ptrtoint i32* [[TMP9]] to i64
// CHECK2-NEXT: [[TMP11:%.*]] = ptrtoint i32* [[ARRAYIDX]] to i64
// CHECK2-NEXT: [[TMP12:%.*]] = sub i64 [[TMP10]], [[TMP11]]
// CHECK2-NEXT: [[TMP13:%.*]] = sdiv exact i64 [[TMP12]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK2-NEXT: [[TMP14:%.*]] = bitcast [3 x i32]* [[OUTPUT4]] to i32*
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr i32, i32* [[TMP14]], i64 [[TMP13]]
// CHECK2-NEXT: store i32* [[TMP15]], i32** [[_TMP5]], align 4
// CHECK2-NEXT: [[RHS_BEGIN:%.*]] = bitcast [3 x i32]* [[OUTPUT4]] to i32*
// CHECK2-NEXT: [[TMP16:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, i32* [[TMP16]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP17]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[CMP7:%.*]] = icmp sgt i32 [[TMP18]], [[TMP19]]
// CHECK2-NEXT: br i1 [[CMP7]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2: cond.true:
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: br label [[COND_END:%.*]]
// CHECK2: cond.false:
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ [[TMP20]], [[COND_TRUE]] ], [ [[TMP21]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP22]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP23]], [[TMP24]]
// CHECK2-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP25]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[I6]], align 4
// CHECK2-NEXT: [[TMP26:%.*]] = load i32*, i32** [[INPUT_ADDR]], align 4
// CHECK2-NEXT: [[TMP27:%.*]] = load i32, i32* [[I6]], align 4
// CHECK2-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i32, i32* [[TMP26]], i32 [[TMP27]]
// CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[ARRAYIDX9]], align 4
// CHECK2-NEXT: [[TMP29:%.*]] = load i32*, i32** [[_TMP5]], align 4
// CHECK2-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, i32* [[TMP29]], i32 0
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, i32* [[ARRAYIDX10]], align 4
// CHECK2-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP30]], [[TMP28]]
// CHECK2-NEXT: store i32 [[ADD11]], i32* [[ARRAYIDX10]], align 4
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP31]], 1
// CHECK2-NEXT: store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit:
// CHECK2-NEXT: [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP33]])
// CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK2-NEXT: [[TMP35:%.*]] = bitcast i32* [[RHS_BEGIN]] to i8*
// CHECK2-NEXT: store i8* [[TMP35]], i8** [[TMP34]], align 4
// CHECK2-NEXT: [[TMP36:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP37:%.*]] = load i32, i32* [[TMP36]], align 4
// CHECK2-NEXT: [[TMP38:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT: [[TMP39:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP37]], i32 1, i32 4, i8* [[TMP38]], void (i8*, i8*)* @.omp.reduction.reduction_func.5, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: switch i32 [[TMP39]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT: ]
// CHECK2: .omp.reduction.case1:
// CHECK2-NEXT: [[TMP40:%.*]] = getelementptr i32, i32* [[ARRAYIDX]], i32 3
// CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[ARRAYIDX]], [[TMP40]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE17:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2: omp.arraycpy.body:
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST13:%.*]] = phi i32* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT15:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[TMP41:%.*]] = load i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST13]], align 4
// CHECK2-NEXT: [[TMP42:%.*]] = load i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK2-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP41]], [[TMP42]]
// CHECK2-NEXT: store i32 [[ADD14]], i32* [[OMP_ARRAYCPY_DESTELEMENTPAST13]], align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT15]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST13]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE16:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT15]], [[TMP40]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE16]], label [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2: omp.arraycpy.done17:
// CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP37]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.case2:
// CHECK2-NEXT: [[TMP43:%.*]] = getelementptr i32, i32* [[ARRAYIDX]], i32 3
// CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY18:%.*]] = icmp eq i32* [[ARRAYIDX]], [[TMP43]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY18]], label [[OMP_ARRAYCPY_DONE25:%.*]], label [[OMP_ARRAYCPY_BODY19:%.*]]
// CHECK2: omp.arraycpy.body19:
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST20:%.*]] = phi i32* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT23:%.*]], [[OMP_ARRAYCPY_BODY19]] ]
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST21:%.*]] = phi i32* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT22:%.*]], [[OMP_ARRAYCPY_BODY19]] ]
// CHECK2-NEXT: [[TMP44:%.*]] = load i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST20]], align 4
// CHECK2-NEXT: [[TMP45:%.*]] = atomicrmw add i32* [[OMP_ARRAYCPY_DESTELEMENTPAST21]], i32 [[TMP44]] monotonic, align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT22]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST21]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT23]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST20]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE24:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT22]], [[TMP43]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE24]], label [[OMP_ARRAYCPY_DONE25]], label [[OMP_ARRAYCPY_BODY19]]
// CHECK2: omp.arraycpy.done25:
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.default:
// CHECK2-NEXT: br label [[OMP_PRECOND_END]]
// CHECK2: omp.precond.end:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 4
// CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i32 0, i32 0
// CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr i32, i32* [[TMP11]], i32 3
// CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[TMP11]], [[TMP12]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2: omp.arraycpy.body:
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2: omp.arraycpy.done2:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 4
// CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i32 0, i32 0
// CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr i32, i32* [[TMP11]], i32 3
// CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[TMP11]], [[TMP12]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2: omp.arraycpy.body:
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2: omp.arraycpy.done2:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l78
// CHECK2-SAME: (i32 [[SIZE:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR5:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK2-NEXT: [[SIZE_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB4]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i32 [[TMP2]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[SIZE:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK2-NEXT: [[A2:%.*]] = alloca [2 x i32], align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i32 0, i32 0
// CHECK2-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i32 0, i32 1
// CHECK2-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[A2]], i32 0, i32 0
// CHECK2-NEXT: [[TMP1:%.*]] = getelementptr i32, i32* [[ARRAY_BEGIN]], i32 2
// CHECK2-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i32* [[ARRAY_BEGIN]], [[TMP1]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2: omp.arrayinit.body:
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT: store i32 0, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP1]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2: omp.arrayinit.done:
// CHECK2-NEXT: [[TMP2:%.*]] = bitcast [10 x i32]* [[TMP0]] to i32*
// CHECK2-NEXT: [[TMP3:%.*]] = ptrtoint i32* [[TMP2]] to i64
// CHECK2-NEXT: [[TMP4:%.*]] = ptrtoint i32* [[ARRAYIDX]] to i64
// CHECK2-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK2-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK2-NEXT: [[TMP7:%.*]] = bitcast [2 x i32]* [[A2]] to i32*
// CHECK2-NEXT: [[TMP8:%.*]] = getelementptr i32, i32* [[TMP7]], i64 [[TMP6]]
// CHECK2-NEXT: [[TMP9:%.*]] = bitcast i32* [[TMP8]] to [10 x i32]*
// CHECK2-NEXT: [[RHS_BEGIN:%.*]] = bitcast [2 x i32]* [[A2]] to i32*
// CHECK2-NEXT: store i32 0, i32* [[I]], align 4
// CHECK2-NEXT: br label [[FOR_COND:%.*]]
// CHECK2: for.cond:
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP10]], [[TMP11]]
// CHECK2-NEXT: br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// CHECK2: for.body:
// CHECK2-NEXT: br label [[FOR_INC:%.*]]
// CHECK2: for.inc:
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK2-NEXT: store i32 [[INC]], i32* [[I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK2-NEXT: [[TMP14:%.*]] = bitcast i32* [[RHS_BEGIN]] to i8*
// CHECK2-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 4
// CHECK2-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4
// CHECK2-NEXT: [[TMP17:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT: [[TMP18:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP16]], i32 1, i32 4, i8* [[TMP17]], void (i8*, i8*)* @.omp.reduction.reduction_func.10, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: switch i32 [[TMP18]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT: ]
// CHECK2: .omp.reduction.case1:
// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr i32, i32* [[ARRAYIDX]], i32 2
// CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[ARRAYIDX]], [[TMP19]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2: omp.arraycpy.body:
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi i32* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP19]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2: omp.arraycpy.done6:
// CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP16]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.case2:
// CHECK2-NEXT: [[TMP22:%.*]] = getelementptr i32, i32* [[ARRAYIDX]], i32 2
// CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq i32* [[ARRAYIDX]], [[TMP22]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
// CHECK2: omp.arraycpy.body8:
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi i32* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi i32* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK2-NEXT: [[TMP23:%.*]] = load i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
// CHECK2-NEXT: [[TMP24:%.*]] = atomicrmw add i32* [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP23]] monotonic, align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP22]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
// CHECK2: omp.arraycpy.done14:
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.default:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.10
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 4
// CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i32 0, i32 0
// CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr i32, i32* [[TMP11]], i32 2
// CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[TMP11]], [[TMP12]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2: omp.arraycpy.body:
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2: omp.arraycpy.done2:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3sumPiiS__l81
// CHECK2-SAME: (i32 [[SIZE:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR5]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK2-NEXT: [[SIZE_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[SIZE_CASTED]], align 4
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB4]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[SIZE:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[SIZE_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK2-NEXT: [[A1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK2-NEXT: store i32 [[SIZE]], i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i32 0, i32 3
// CHECK2-NEXT: store i32 0, i32* [[A1]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = bitcast [10 x i32]* [[TMP0]] to i32*
// CHECK2-NEXT: [[TMP2:%.*]] = ptrtoint i32* [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = ptrtoint i32* [[ARRAYIDX]] to i64
// CHECK2-NEXT: [[TMP4:%.*]] = sub i64 [[TMP2]], [[TMP3]]
// CHECK2-NEXT: [[TMP5:%.*]] = sdiv exact i64 [[TMP4]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr i32, i32* [[A1]], i64 [[TMP5]]
// CHECK2-NEXT: [[TMP7:%.*]] = bitcast i32* [[TMP6]] to [10 x i32]*
// CHECK2-NEXT: store i32 0, i32* [[I]], align 4
// CHECK2-NEXT: br label [[FOR_COND:%.*]]
// CHECK2: for.cond:
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[SIZE_ADDR]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP8]], [[TMP9]]
// CHECK2-NEXT: br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// CHECK2: for.body:
// CHECK2-NEXT: br label [[FOR_INC:%.*]]
// CHECK2: for.inc:
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK2-NEXT: store i32 [[INC]], i32* [[I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK2-NEXT: [[TMP12:%.*]] = bitcast i32* [[A1]] to i8*
// CHECK2-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 4
// CHECK2-NEXT: [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[TMP13]], align 4
// CHECK2-NEXT: [[TMP15:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], i32 1, i32 4, i8* [[TMP15]], void (i8*, i8*)* @.omp.reduction.reduction_func.14, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: switch i32 [[TMP16]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT: ]
// CHECK2: .omp.reduction.case1:
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, i32* [[A1]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[ARRAYIDX]], align 4
// CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.case2:
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[A1]], align 4
// CHECK2-NEXT: [[TMP20:%.*]] = atomicrmw add i32* [[ARRAYIDX]], i32 [[TMP19]] monotonic, align 4
// CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2: .omp.reduction.default:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.14
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 4
// CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 4
// CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i32 0, i32 0
// CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@main
// CHECK2-SAME: () #[[ATTR6:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SIZE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[ARRAY:%.*]] = alloca i32*, align 4
// CHECK2-NEXT: [[RESULT:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK2-NEXT: store i32 100, i32* [[SIZE]], align 4
// CHECK2-NEXT: [[CALL:%.*]] = call noalias nonnull i8* @_Znaj(i32 400) #[[ATTR9:[0-9]+]]
// CHECK2-NEXT: [[TMP0:%.*]] = bitcast i8* [[CALL]] to i32*
// CHECK2-NEXT: store i32* [[TMP0]], i32** [[ARRAY]], align 4
// CHECK2-NEXT: store i32 0, i32* [[RESULT]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[ARRAY]], align 4
// CHECK2-NEXT: call void @_Z3sumPiiS_(i32* [[TMP1]], i32 100, i32* [[RESULT]])
// CHECK2-NEXT: ret i32 0
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK2-SAME: () #[[ATTR8:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK2-NEXT: ret void
//

clang/test/OpenMP/target_codegen.cpp

Show First 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// code and have mapped arguments, and only 6 have all-constant map sizes. // code and have mapped arguments, and only 6 have all-constant map sizes.
// CHECK-DAG: [[SIZET:@.+]] = private unnamed_addr constant [2 x i64] [i64 0, i64 4] // CHECK-DAG: [[SIZET:@.+]] = private unnamed_addr constant [2 x i64] [i64 0, i64 4]
// CHECK-DAG: [[MAPT:@.+]] = private unnamed_addr constant [2 x i64] [i64 544, i64 800] // CHECK-DAG: [[MAPT:@.+]] = private unnamed_addr constant [2 x i64] [i64 544, i64 800]
// CHECK-DAG: [[SIZET2:@.+]] = private unnamed_addr constant [1 x i{{32|64}}] [i64 2] // CHECK-DAG: [[SIZET2:@.+]] = private unnamed_addr constant [1 x i{{32|64}}] [i64 2]
// CHECK-DAG: [[MAPT2:@.+]] = private unnamed_addr constant [1 x i64] [i64 800] // CHECK-DAG: [[MAPT2:@.+]] = private unnamed_addr constant [1 x i64] [i64 800]
// CHECK-DAG: [[SIZET3:@.+]] = private unnamed_addr constant [2 x i64] [i64 4, i64 2] // CHECK-DAG: [[SIZET3:@.+]] = private unnamed_addr constant [2 x i64] [i64 4, i64 2]
// CHECK-DAG: [[MAPT3:@.+]] = private unnamed_addr constant [2 x i64] [i64 800, i64 800] // CHECK-DAG: [[MAPT3:@.+]] = private unnamed_addr constant [2 x i64] [i64 800, i64 800]
// CHECK-DAG: [[SIZET4:@.+]] = private unnamed_addr global [9 x i64] [i64 4, i64 40, i64 {{4|8}}, i64 0, i64 400, i64 {{4|8}}, i64 {{4|8}}, i64 0, i64 {{12|16}}]
// CHECK-DAG: [[MAPT4:@.+]] = private unnamed_addr constant [9 x i64] [i64 800, i64 547, i64 800, i64 547, i64 547, i64 800, i64 800, i64 547, i64 547] // CHECK-DAG: [[MAPT4:@.+]] = private unnamed_addr constant [9 x i64] [i64 800, i64 547, i64 800, i64 547, i64 547, i64 800, i64 800, i64 547, i64 547]
// CHECK-DAG: [[SIZET5:@.+]] = private unnamed_addr constant [3 x i64] [i64 4, i64 2, i64 40] // CHECK-DAG: [[SIZET5:@.+]] = private unnamed_addr constant [3 x i64] [i64 4, i64 2, i64 40]
// CHECK-DAG: [[MAPT5:@.+]] = private unnamed_addr constant [3 x i64] [i64 800, i64 800, i64 547] // CHECK-DAG: [[MAPT5:@.+]] = private unnamed_addr constant [3 x i64] [i64 800, i64 800, i64 547]
// CHECK-DAG: [[SIZET6:@.+]] = private unnamed_addr constant [4 x i64] [i64 4, i64 2, i64 1, i64 40] // CHECK-DAG: [[SIZET6:@.+]] = private unnamed_addr constant [4 x i64] [i64 4, i64 2, i64 1, i64 40]
// CHECK-DAG: [[MAPT6:@.+]] = private unnamed_addr constant [4 x i64] [i64 800, i64 800, i64 800, i64 547] // CHECK-DAG: [[MAPT6:@.+]] = private unnamed_addr constant [4 x i64] [i64 800, i64 800, i64 800, i64 547]
// CHECK-DAG: [[SIZET7:@.+]] = private unnamed_addr global [5 x i64] [i64 8, i64 4, i64 {{4|8}}, i64 {{4|8}}, i64 0]
// CHECK-DAG: [[MAPT7:@.+]] = private unnamed_addr constant [5 x i64] [i64 547, i64 800, i64 800, i64 800, i64 547] // CHECK-DAG: [[MAPT7:@.+]] = private unnamed_addr constant [5 x i64] [i64 547, i64 800, i64 800, i64 800, i64 547]
// CHECK-DAG: [[SIZET9:@.+]] = private unnamed_addr constant [1 x i64] [i64 12] // CHECK-DAG: [[SIZET9:@.+]] = private unnamed_addr constant [1 x i64] [i64 12]
// CHECK-DAG: [[MAPT10:@.+]] = private unnamed_addr constant [1 x i64] [i64 35] // CHECK-DAG: [[MAPT10:@.+]] = private unnamed_addr constant [1 x i64] [i64 35]
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
▲ Show 20 Lines • Show All 175 Lines • ▼ Show 20 Linesint foo(int n) {
// CHECK-64: [[BNSIZE:%.+]] = mul nuw i64 [[VLA0:%.+]], 4 // CHECK-64: [[BNSIZE:%.+]] = mul nuw i64 [[VLA0:%.+]], 4
// CHECK-32: [[BNSZSIZE:%.+]] = mul nuw i32 [[VLA0:%.+]], 4 // CHECK-32: [[BNSZSIZE:%.+]] = mul nuw i32 [[VLA0:%.+]], 4
// CHECK-32: [[BNSIZE:%.+]] = sext i32 [[BNSZSIZE]] to i64 // CHECK-32: [[BNSIZE:%.+]] = sext i32 [[BNSZSIZE]] to i64
// CHECK: [[CNELEMSIZE2:%.+]] = mul nuw i[[SZ]] 5, [[VLA1:%.+]] // CHECK: [[CNELEMSIZE2:%.+]] = mul nuw i[[SZ]] 5, [[VLA1:%.+]]
// CHECK-64: [[CNSIZE:%.+]] = mul nuw i64 [[CNELEMSIZE2]], 8 // CHECK-64: [[CNSIZE:%.+]] = mul nuw i64 [[CNELEMSIZE2]], 8
// CHECK-32: [[CNSZSIZE:%.+]] = mul nuw i32 [[CNELEMSIZE2]], 8 // CHECK-32: [[CNSZSIZE:%.+]] = mul nuw i32 [[CNELEMSIZE2]], 8
// CHECK-32: [[CNSIZE:%.+]] = sext i32 [[CNSZSIZE]] to i64 // CHECK-32: [[CNSIZE:%.+]] = sext i32 [[CNSZSIZE]] to i64
// CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 9, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* [[SR:%[^,]+]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[MAPT4]], i32 0, i32 0), i8** null, i8** null) // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 9, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[SIZET4]], i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[MAPT4]], i32 0, i32 0), i8** null, i8** null)
// CHECK-DAG: [[BPR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP:%[^,]+]], i32 0, i32 0 // CHECK-DAG: [[BPR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP:%[^,]+]], i32 0, i32 0
// CHECK-DAG: [[PR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P:%[^,]+]], i32 0, i32 0 // CHECK-DAG: [[PR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P:%[^,]+]], i32 0, i32 0
// CHECK-DAG: [[SR]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S:%[^,]+]], i32 0, i32 0
// CHECK-DAG: [[SADDR0:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX0:0]] // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX0:0]]
// CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX0]]
// CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX0]] // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX0]]
// CHECK-DAG: [[SADDR1:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX1:1]] // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX1:1]]
// CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX1]]
// CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX1]] // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX1]]
// CHECK-DAG: [[SADDR2:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX2:2]] // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX2:2]]
// CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX2]]
// CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX2]] // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX2]]
// CHECK-DAG: [[SADDR3:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX3:3]] // CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX3:3]]
// CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX3]]
// CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX3]] // CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX3]]
// CHECK-DAG: [[SADDR4:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX4:4]] // CHECK-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX4:4]]
// CHECK-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX4]]
// CHECK-DAG: [[PADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX4]] // CHECK-DAG: [[PADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX4]]
// CHECK-DAG: [[SADDR5:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX5:5]] // CHECK-DAG: [[BPADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX5:5]]
// CHECK-DAG: [[BPADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX5]]
// CHECK-DAG: [[PADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX5]] // CHECK-DAG: [[PADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX5]]
// CHECK-DAG: [[SADDR6:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX6:6]] // CHECK-DAG: [[BPADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX6:6]]
// CHECK-DAG: [[BPADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX6]]
// CHECK-DAG: [[PADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX6]] // CHECK-DAG: [[PADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX6]]
// CHECK-DAG: [[SADDR7:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX7:7]] // CHECK-DAG: [[BPADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX7:7]]
// CHECK-DAG: [[BPADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX7]]
// CHECK-DAG: [[PADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX7]] // CHECK-DAG: [[PADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX7]]
// CHECK-DAG: [[SADDR8:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX8:8]] // CHECK-DAG: [[BPADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX8:8]]
// CHECK-DAG: [[BPADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX8]]
// CHECK-DAG: [[PADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX8]] // CHECK-DAG: [[PADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX8]]
// The names below are not necessarily consistent with the names used for the // The names below are not necessarily consistent with the names used for the
// addresses above as some are repeated. // addresses above as some are repeated.
// CHECK-DAG: [[CBPADDR2:%.+]] = bitcast i8** [[BPADDR2]] to i[[SZ]]* // CHECK-DAG: [[CBPADDR2:%.+]] = bitcast i8** [[BPADDR2]] to i[[SZ]]*
// CHECK-DAG: [[CPADDR2:%.+]] = bitcast i8** [[PADDR2]] to i[[SZ]]* // CHECK-DAG: [[CPADDR2:%.+]] = bitcast i8** [[PADDR2]] to i[[SZ]]*
// CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CBPADDR2]] // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CBPADDR2]]
// CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CPADDR2]] // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CPADDR2]]
// CHECK-DAG: store i64 {{4|8}}, i64* [[SADDR2]]
// CHECK-DAG: [[CBPADDR6:%.+]] = bitcast i8** [[BPADDR6]] to i[[SZ]]* // CHECK-DAG: [[CBPADDR6:%.+]] = bitcast i8** [[BPADDR6]] to i[[SZ]]*
// CHECK-DAG: [[CPADDR6:%.+]] = bitcast i8** [[PADDR6]] to i[[SZ]]* // CHECK-DAG: [[CPADDR6:%.+]] = bitcast i8** [[PADDR6]] to i[[SZ]]*
// CHECK-DAG: store i[[SZ]] [[VLA1]], i[[SZ]]* [[CBPADDR6]] // CHECK-DAG: store i[[SZ]] [[VLA1]], i[[SZ]]* [[CBPADDR6]]
// CHECK-DAG: store i[[SZ]] [[VLA1]], i[[SZ]]* [[CPADDR6]] // CHECK-DAG: store i[[SZ]] [[VLA1]], i[[SZ]]* [[CPADDR6]]
// CHECK-DAG: store i64 {{4|8}}, i64* [[SADDR6]]
// CHECK-DAG: [[CBPADDR5:%.+]] = bitcast i8** [[BPADDR5]] to i[[SZ]]* // CHECK-DAG: [[CBPADDR5:%.+]] = bitcast i8** [[BPADDR5]] to i[[SZ]]*
// CHECK-DAG: [[CPADDR5:%.+]] = bitcast i8** [[PADDR5]] to i[[SZ]]* // CHECK-DAG: [[CPADDR5:%.+]] = bitcast i8** [[PADDR5]] to i[[SZ]]*
// CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CBPADDR5]] // CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CBPADDR5]]
// CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CPADDR5]] // CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CPADDR5]]
// CHECK-DAG: store i64 {{4|8}}, i64* [[SADDR5]]
// CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to i[[SZ]]* // CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to i[[SZ]]*
// CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to i[[SZ]]* // CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to i[[SZ]]*
// CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CBPADDR0]] // CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CBPADDR0]]
// CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CPADDR0]] // CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CPADDR0]]
// CHECK-DAG: store i64 4, i64* [[SADDR0]]
// CHECK-DAG: [[CBPADDR1:%.+]] = bitcast i8** [[BPADDR1]] to [10 x float]** // CHECK-DAG: [[CBPADDR1:%.+]] = bitcast i8** [[BPADDR1]] to [10 x float]**
// CHECK-DAG: [[CPADDR1:%.+]] = bitcast i8** [[PADDR1]] to [10 x float]** // CHECK-DAG: [[CPADDR1:%.+]] = bitcast i8** [[PADDR1]] to [10 x float]**
// CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CBPADDR1]] // CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CBPADDR1]]
// CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CPADDR1]] // CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CPADDR1]]
// CHECK-DAG: store i64 40, i64* [[SADDR1]]
// CHECK-DAG: [[CBPADDR3:%.+]] = bitcast i8** [[BPADDR3]] to float** // CHECK-DAG: [[CBPADDR3:%.+]] = bitcast i8** [[BPADDR3]] to float**
// CHECK-DAG: [[CPADDR3:%.+]] = bitcast i8** [[PADDR3]] to float** // CHECK-DAG: [[CPADDR3:%.+]] = bitcast i8** [[PADDR3]] to float**
// CHECK-DAG: store float* %{{.+}}, float** [[CBPADDR3]] // CHECK-DAG: store float* %{{.+}}, float** [[CBPADDR3]]
// CHECK-DAG: store float* %{{.+}}, float** [[CPADDR3]] // CHECK-DAG: store float* %{{.+}}, float** [[CPADDR3]]
// CHECK-DAG: store i64 [[BNSIZE]], i64* [[SADDR3]] // CHECK-DAG: store i64 [[BNSIZE]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[SIZET4]], i32 0, i32 [[IDX3]])
// CHECK-DAG: [[CBPADDR4:%.+]] = bitcast i8** [[BPADDR4]] to [5 x [10 x double]]** // CHECK-DAG: [[CBPADDR4:%.+]] = bitcast i8** [[BPADDR4]] to [5 x [10 x double]]**
// CHECK-DAG: [[CPADDR4:%.+]] = bitcast i8** [[PADDR4]] to [5 x [10 x double]]** // CHECK-DAG: [[CPADDR4:%.+]] = bitcast i8** [[PADDR4]] to [5 x [10 x double]]**
// CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CBPADDR4]] // CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CBPADDR4]]
// CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CPADDR4]] // CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CPADDR4]]
// CHECK-DAG: store i64 400, i64* [[SADDR4]]
// CHECK-DAG: [[CBPADDR7:%.+]] = bitcast i8** [[BPADDR7]] to double** // CHECK-DAG: [[CBPADDR7:%.+]] = bitcast i8** [[BPADDR7]] to double**
// CHECK-DAG: [[CPADDR7:%.+]] = bitcast i8** [[PADDR7]] to double** // CHECK-DAG: [[CPADDR7:%.+]] = bitcast i8** [[PADDR7]] to double**
// CHECK-DAG: store double* %{{.+}}, double** [[CBPADDR7]] // CHECK-DAG: store double* %{{.+}}, double** [[CBPADDR7]]
// CHECK-DAG: store double* %{{.+}}, double** [[CPADDR7]] // CHECK-DAG: store double* %{{.+}}, double** [[CPADDR7]]
// CHECK-DAG: store i64 [[CNSIZE]], i64* [[SADDR7]] // CHECK-DAG: store i64 [[CNSIZE]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[SIZET4]], i32 0, i32 [[IDX7]])
// CHECK-DAG: [[CBPADDR8:%.+]] = bitcast i8** [[BPADDR8]] to [[TT]]** // CHECK-DAG: [[CBPADDR8:%.+]] = bitcast i8** [[BPADDR8]] to [[TT]]**
// CHECK-DAG: [[CPADDR8:%.+]] = bitcast i8** [[PADDR8]] to [[TT]]** // CHECK-DAG: [[CPADDR8:%.+]] = bitcast i8** [[PADDR8]] to [[TT]]**
// CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CBPADDR8]] // CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CBPADDR8]]
// CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CPADDR8]] // CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CPADDR8]]
// CHECK-DAG: store i64 {{12|16}}, i64* [[SADDR8]]
// CHECK: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 // CHECK: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0
// CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]] // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]]
// CHECK: [[FAIL]] // CHECK: [[FAIL]]
// CHECK: call void [[HVT4:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) // CHECK: call void [[HVT4:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}})
// CHECK-NEXT: br label %[[END]] // CHECK-NEXT: br label %[[END]]
// CHECK: [[END]] // CHECK: [[END]]
// CHECK-NEXT: br label %[[IFEND:.+]] // CHECK-NEXT: br label %[[IFEND:.+]]
▲ Show 20 Lines • Show All 204 Lines • ▼ Show 20 Lines
// CHECK: br i1 [[IF]], label %[[TRY:[^,]+]], label %[[IFELSE:[^,]+]] // CHECK: br i1 [[IF]], label %[[TRY:[^,]+]], label %[[IFELSE:[^,]+]]
// CHECK: [[TRY]] // CHECK: [[TRY]]
// We capture 2 VLA sizes in this target region // We capture 2 VLA sizes in this target region
// CHECK: [[CELEMSIZE2:%.+]] = mul nuw i[[SZ]] 2, [[VLA0:%.+]] // CHECK: [[CELEMSIZE2:%.+]] = mul nuw i[[SZ]] 2, [[VLA0:%.+]]
// CHECK-64: [[CSIZE:%.+]] = mul nuw i64 [[CELEMSIZE2]], 2 // CHECK-64: [[CSIZE:%.+]] = mul nuw i64 [[CELEMSIZE2]], 2
// CHECK-32: [[CSZSIZE:%.+]] = mul nuw i32 [[CELEMSIZE2]], 2 // CHECK-32: [[CSZSIZE:%.+]] = mul nuw i32 [[CELEMSIZE2]], 2
// CHECK-32: [[CSIZE:%.+]] = sext i32 [[CSZSIZE]] to i64 // CHECK-32: [[CSIZE:%.+]] = sext i32 [[CSZSIZE]] to i64
// CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 5, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* [[SR:%[^,]+]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPT7]], i32 0, i32 0), i8** null, i8** null) // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 5, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZET7]], i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPT7]], i32 0, i32 0), i8** null, i8** null)
// CHECK-DAG: [[BPR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP:%.+]], i32 0, i32 0 // CHECK-DAG: [[BPR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP:%.+]], i32 0, i32 0
// CHECK-DAG: [[PR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P:%.+]], i32 0, i32 0 // CHECK-DAG: [[PR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P:%.+]], i32 0, i32 0
// CHECK-DAG: [[SR]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S:%.+]], i32 0, i32 0 // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 [[IDX0:0]]
// CHECK-DAG: [[SADDR0:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S]], i32 0, i32 [[IDX0:0]]
// CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 [[IDX0]]
// CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 [[IDX0]] // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 [[IDX0]]
// CHECK-DAG: [[SADDR1:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S]], i32 0, i32 [[IDX1:1]] // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 [[IDX1:1]]
// CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 [[IDX1]]
// CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 [[IDX1]] // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 [[IDX1]]
// CHECK-DAG: [[SADDR2:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S]], i32 0, i32 [[IDX2:2]] // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 [[IDX2:2]]
// CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 [[IDX2]]
// CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 [[IDX2]] // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 [[IDX2]]
// CHECK-DAG: [[SADDR3:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S]], i32 0, i32 [[IDX3:3]] // CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 [[IDX3:3]]
// CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 [[IDX3]]
// CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 [[IDX3]] // CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 [[IDX3]]
// CHECK-DAG: [[SADDR4:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S]], i32 0, i32 [[IDX4:4]] // CHECK-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 [[IDX4:4]]
// CHECK-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 0, i32 [[IDX4]]
// CHECK-DAG: [[PADDR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 [[IDX4]] // CHECK-DAG: [[PADDR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 0, i32 [[IDX4]]
// The names below are not necessarily consistent with the names used for the // The names below are not necessarily consistent with the names used for the
// addresses above as some are repeated. // addresses above as some are repeated.
// CHECK-DAG: [[CBPADDR4:%.+]] = bitcast i8** [[BPADDR4]] to i16** // CHECK-DAG: [[CBPADDR4:%.+]] = bitcast i8** [[BPADDR4]] to i16**
// CHECK-DAG: [[CPADDR4:%.+]] = bitcast i8** [[PADDR4]] to i16** // CHECK-DAG: [[CPADDR4:%.+]] = bitcast i8** [[PADDR4]] to i16**
// CHECK-DAG: store i16* %{{.+}}, i16** [[CBPADDR4]] // CHECK-DAG: store i16* %{{.+}}, i16** [[CBPADDR4]]
// CHECK-DAG: store i16* %{{.+}}, i16** [[CPADDR4]] // CHECK-DAG: store i16* %{{.+}}, i16** [[CPADDR4]]
// CHECK-DAG: store i64 [[CSIZE]], i64* [[SADDR4]] // CHECK-DAG: store i64 [[CSIZE]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZET7]], i32 0, i32 [[IDX4]])
// CHECK-DAG: [[CBPADDR3:%.+]] = bitcast i8** [[BPADDR3]] to i[[SZ]]* // CHECK-DAG: [[CBPADDR3:%.+]] = bitcast i8** [[BPADDR3]] to i[[SZ]]*
// CHECK-DAG: [[CPADDR3:%.+]] = bitcast i8** [[PADDR3]] to i[[SZ]]* // CHECK-DAG: [[CPADDR3:%.+]] = bitcast i8** [[PADDR3]] to i[[SZ]]*
// CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CBPADDR3]] // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CBPADDR3]]
// CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CPADDR3]] // CHECK-DAG: store i[[SZ]] [[VLA0]], i[[SZ]]* [[CPADDR3]]
// CHECK-DAG: store i64 {{4|8}}, i64* [[SADDR3]]
// CHECK-DAG: [[CBPADDR2:%.+]] = bitcast i8** [[BPADDR2]] to i[[SZ]]* // CHECK-DAG: [[CBPADDR2:%.+]] = bitcast i8** [[BPADDR2]] to i[[SZ]]*
// CHECK-DAG: [[CPADDR2:%.+]] = bitcast i8** [[PADDR2]] to i[[SZ]]* // CHECK-DAG: [[CPADDR2:%.+]] = bitcast i8** [[PADDR2]] to i[[SZ]]*
// CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CBPADDR2]] // CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CBPADDR2]]
// CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CPADDR2]] // CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CPADDR2]]
// CHECK-DAG: store i64 {{4|8}}, i64* [[SADDR2]]
// CHECK-DAG: [[CBPADDR1:%.+]] = bitcast i8** [[BPADDR1]] to i[[SZ]]* // CHECK-DAG: [[CBPADDR1:%.+]] = bitcast i8** [[BPADDR1]] to i[[SZ]]*
// CHECK-DAG: [[CPADDR1:%.+]] = bitcast i8** [[PADDR1]] to i[[SZ]]* // CHECK-DAG: [[CPADDR1:%.+]] = bitcast i8** [[PADDR1]] to i[[SZ]]*
// CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CBPADDR1]] // CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CBPADDR1]]
// CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CPADDR1]] // CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CPADDR1]]
// CHECK-DAG: store i64 4, i64* [[SADDR1]]
// CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to [[S1]]** // CHECK-DAG: [[CBPADDR0:%.+]] = bitcast i8** [[BPADDR0]] to [[S1]]**
// CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to double** // CHECK-DAG: [[CPADDR0:%.+]] = bitcast i8** [[PADDR0]] to double**
// CHECK-DAG: store [[S1]]* [[THIS:%.+]], [[S1]]** [[CBPADDR0]] // CHECK-DAG: store [[S1]]* [[THIS:%.+]], [[S1]]** [[CBPADDR0]]
// CHECK-DAG: store double* [[A:%.+]], double** [[CPADDR0]] // CHECK-DAG: store double* [[A:%.+]], double** [[CPADDR0]]
// CHECK-DAG: store i64 8, i64* [[SADDR0]]
// CHECK: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 // CHECK: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0
// CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]] // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]]
// CHECK: [[FAIL]] // CHECK: [[FAIL]]
// CHECK: call void [[HVT7:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) // CHECK: call void [[HVT7:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}})
// CHECK-NEXT: br label %[[END]] // CHECK-NEXT: br label %[[END]]
// CHECK: [[END]] // CHECK: [[END]]
// CHECK-NEXT: br label %[[IFEND:.+]] // CHECK-NEXT: br label %[[IFEND:.+]]
▲ Show 20 Lines • Show All 259 Lines • Show Last 20 Lines

clang/test/OpenMP/target_data_codegen.cpp

Show All 31 Lines
// CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 800] // CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 800]
// CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] [i64 2] // CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] [i64 2]
// CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i[[sz]]] [i[[sz]] 4] // CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i[[sz]]] [i[[sz]] 4]
// CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 1] // CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 1]
// CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 5] // CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 5]
// CK1: [[SIZE04:@.+]] = {{.+}}constant [2 x i64] [i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64 24] // CK1: [[SIZE04:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673] // CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673]
// CK1: [[MTYPE05:@.+]] = {{.+}}constant [1 x i64] [i64 1025] // CK1: [[MTYPE05:@.+]] = {{.+}}constant [1 x i64] [i64 1025]
// CK1: [[MTYPE06:@.+]] = {{.+}}constant [1 x i64] [i64 1029] // CK1: [[MTYPE06:@.+]] = {{.+}}constant [1 x i64] [i64 1029]
// CK1-LABEL: _Z3fooi // CK1-LABEL: _Z3fooi
void foo(int arg) { void foo(int arg) {
▲ Show 20 Lines • Show All 96 Lines • ▼ Show 20 Lines// CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P]]
// CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK1-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK1-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK1-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK1-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK1-DAG: store [[ST]]* @gb, [[ST]]** [[CBP0]] // CK1-DAG: store [[ST]]* @gb, [[ST]]** [[CBP0]]
// CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CP0]] // CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CP0]]
// CK1-DAG: store i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE04]], i32 0, i32 0),
// CK1-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK1-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK1-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK1-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK1-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK1-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK1-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK1-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CBP1]] // CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CBP1]]
// CK1-DAG: store double* [[SEC1:%.+]], double** [[CP1]] // CK1-DAG: store double* [[SEC1:%.+]], double** [[CP1]]
// CK1-DAG: [[SEC1]] = getelementptr inbounds {{.+}}double* [[SEC11:%[^,]+]], i{{.+}} 0 // CK1-DAG: [[SEC1]] = getelementptr inbounds {{.+}}double* [[SEC11:%[^,]+]], i{{.+}} 0
▲ Show 20 Lines • Show All 191 Lines • ▼ Show 20 Linesstruct ST {
T foo(T arg) { T foo(T arg) {
// Region 00 // Region 00
#pragma omp target data map(always, to: b[1:3]) if(a>123) device(arg) #pragma omp target data map(always, to: b[1:3]) if(a>123) device(arg)
{arg++;} {arg++;}
return arg; return arg;
} }
}; };
// CK2: [[SIZE00:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK2: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710677] // CK2: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710677]
// CK2-LABEL: _Z3bari // CK2-LABEL: _Z3bari
int bar(int arg){ int bar(int arg){
ST<int> A; ST<int> A;
return A.foo(arg); return A.foo(arg);
} }
// Region 00 // Region 00
// CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] // CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
// CK2: [[IFTHEN]] // CK2: [[IFTHEN]]
// CK2-DAG: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[sz:64|32]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null, i8** null) // CK2-DAG: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null, i8** null)
// CK2-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64 // CK2-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64
// CK2-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}}, // CK2-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}},
// CK2-DAG: [[GEPBP]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%[^,]+]] // CK2-DAG: [[GEPBP]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%[^,]+]]
// CK2-DAG: [[GEPP]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%[^,]+]] // CK2-DAG: [[GEPP]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%[^,]+]]
// CK2-DAG: [[GEPS]] = getelementptr inbounds [2 x i[[sz]]], [2 x i[[sz]]]* [[S:%[^,]+]]
// CK2-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK2-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK2-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK2-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK2-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK2-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK2-DAG: store [[ST]]* [[VAR0:%.+]], [[ST]]** [[CBP0]] // CK2-DAG: store [[ST]]* [[VAR0:%.+]], [[ST]]** [[CBP0]]
// CK2-DAG: store double** [[SEC0:%.+]], double*** [[CP0]] // CK2-DAG: store double** [[SEC0:%.+]], double*** [[CP0]]
// CK2-DAG: store i[[sz]] {{%.+}}, i[[sz]]* [[S0]] // CK2-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE00]], i32 0, i32 0),
// CK2-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1 // CK2-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1
// CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK2-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK2-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK2-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK2-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK2-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK2-DAG: store double** [[SEC0]], double*** [[CBP1]] // CK2-DAG: store double** [[SEC0]], double*** [[CBP1]]
// CK2-DAG: store double* [[SEC1:%.+]], double** [[CP1]] // CK2-DAG: store double* [[SEC1:%.+]], double** [[CP1]]
// CK2-DAG: store i[[sz]] 24, i[[sz]]* [[S1]]
// CK2-DAG: [[SEC1]] = getelementptr inbounds {{.*}}double* [[SEC11:%[^,]+]], i{{.+}} 1 // CK2-DAG: [[SEC1]] = getelementptr inbounds {{.*}}double* [[SEC11:%[^,]+]], i{{.+}} 1
// CK2-DAG: [[SEC11]] = load double*, double** [[SEC111:%[^,]+]], // CK2-DAG: [[SEC11]] = load double*, double** [[SEC111:%[^,]+]],
// CK2-DAG: [[SEC111]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1 // CK2-DAG: [[SEC111]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1
// CK2: br label %[[IFEND:[^,]+]] // CK2: br label %[[IFEND:[^,]+]]
// CK2: [[IFELSE]] // CK2: [[IFELSE]]
// CK2: br label %[[IFEND]] // CK2: br label %[[IFEND]]
// CK2: [[IFEND]] // CK2: [[IFEND]]
// CK2: %{{.+}} = add nsw i32 %{{[^,]+}}, 1 // CK2: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
// CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] // CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
// CK2: [[IFTHEN]] // CK2: [[IFTHEN]]
// CK2-DAG: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[sz]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null, i8** null) // CK2-DAG: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null, i8** null)
// CK2-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64 // CK2-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64
// CK2-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}}, // CK2-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}},
// CK2-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP]] // CK2-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP]]
// CK2-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P]] // CK2-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P]]
// CK2-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S]]
// CK2: br label %[[IFEND:[^,]+]] // CK2: br label %[[IFEND:[^,]+]]
// CK2: [[IFELSE]] // CK2: [[IFELSE]]
// CK2: br label %[[IFEND]] // CK2: br label %[[IFEND]]
// CK2: [[IFEND]] // CK2: [[IFEND]]
#endif #endif
///==========================================================================/// ///==========================================================================///
// RUN: %clang_cc1 -DCK3 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK3 --check-prefix CK3-64 // RUN: %clang_cc1 -DCK3 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK3 --check-prefix CK3-64
// RUN: %clang_cc1 -DCK3 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK3 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
▲ Show 20 Lines • Show All 47 Lines • ▼ Show 20 Linesstruct STT {
T foo(T arg) { T foo(T arg) {
// Region 00 // Region 00
#pragma omp target data map(always, close to: b[1:3]) if(a>123) device(arg) #pragma omp target data map(always, close to: b[1:3]) if(a>123) device(arg)
{arg++;} {arg++;}
return arg; return arg;
} }
}; };
// CK4: [[SIZE00:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK4: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976711701] // CK4: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976711701]
// CK4-LABEL: _Z3bari // CK4-LABEL: _Z3bari
int bar(int arg){ int bar(int arg){
STT<int> A; STT<int> A;
return A.foo(arg); return A.foo(arg);
} }
// Region 00 // Region 00
// CK4: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] // CK4: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
// CK4: [[IFTHEN]] // CK4: [[IFTHEN]]
// CK4-DAG: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[sz:64|32]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null, i8** null) // CK4-DAG: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null, i8** null)
// CK4-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64 // CK4-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64
// CK4-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}}, // CK4-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}},
// CK4-DAG: [[GEPBP]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%[^,]+]] // CK4-DAG: [[GEPBP]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%[^,]+]]
// CK4-DAG: [[GEPP]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%[^,]+]] // CK4-DAG: [[GEPP]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%[^,]+]]
// CK4-DAG: [[GEPS]] = getelementptr inbounds [2 x i[[sz]]], [2 x i[[sz]]]* [[S:%[^,]+]]
// CK4-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK4-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK4-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK4-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK4-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK4-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[STT]]** // CK4-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[STT]]**
// CK4-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK4-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK4-DAG: store [[STT]]* [[VAR0:%.+]], [[STT]]** [[CBP0]] // CK4-DAG: store [[STT]]* [[VAR0:%.+]], [[STT]]** [[CBP0]]
// CK4-DAG: store double** [[SEC0:%.+]], double*** [[CP0]] // CK4-DAG: store double** [[SEC0:%.+]], double*** [[CP0]]
// CK4-DAG: store i[[sz]] {{%.+}}, i[[sz]]* [[S0]] // CK4-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE00]], i32 0, i32 0),
// CK4-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[STT]]* [[VAR0]], i32 0, i32 1 // CK4-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[STT]]* [[VAR0]], i32 0, i32 1
// CK4-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK4-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK4-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK4-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK4-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK4-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK4-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK4-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK4-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK4-DAG: store double** [[SEC0]], double*** [[CBP1]] // CK4-DAG: store double** [[SEC0]], double*** [[CBP1]]
// CK4-DAG: store double* [[SEC1:%.+]], double** [[CP1]] // CK4-DAG: store double* [[SEC1:%.+]], double** [[CP1]]
// CK4-DAG: store i[[sz]] 24, i[[sz]]* [[S1]]
// CK4-DAG: [[SEC1]] = getelementptr inbounds {{.*}}double* [[SEC11:%[^,]+]], i{{.+}} 1 // CK4-DAG: [[SEC1]] = getelementptr inbounds {{.*}}double* [[SEC11:%[^,]+]], i{{.+}} 1
// CK4-DAG: [[SEC11]] = load double*, double** [[SEC111:%[^,]+]], // CK4-DAG: [[SEC11]] = load double*, double** [[SEC111:%[^,]+]],
// CK4-DAG: [[SEC111]] = getelementptr inbounds {{.*}}[[STT]]* [[VAR0]], i32 0, i32 1 // CK4-DAG: [[SEC111]] = getelementptr inbounds {{.*}}[[STT]]* [[VAR0]], i32 0, i32 1
// CK4: br label %[[IFEND:[^,]+]] // CK4: br label %[[IFEND:[^,]+]]
// CK4: [[IFELSE]] // CK4: [[IFELSE]]
// CK4: br label %[[IFEND]] // CK4: br label %[[IFEND]]
// CK4: [[IFEND]] // CK4: [[IFEND]]
// CK4: %{{.+}} = add nsw i32 %{{[^,]+}}, 1 // CK4: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
// CK4: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] // CK4: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
// CK4: [[IFTHEN]] // CK4: [[IFTHEN]]
// CK4-DAG: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[sz]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null, i8** null) // CK4-DAG: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null, i8** null)
// CK4-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64 // CK4-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64
// CK4-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}}, // CK4-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}},
// CK4-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP]] // CK4-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP]]
// CK4-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P]] // CK4-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P]]
// CK4-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S]]
// CK4: br label %[[IFEND:[^,]+]] // CK4: br label %[[IFEND:[^,]+]]
// CK4: [[IFELSE]] // CK4: [[IFELSE]]
// CK4: br label %[[IFEND]] // CK4: br label %[[IFEND]]
// CK4: [[IFEND]] // CK4: [[IFEND]]
#endif #endif
///==========================================================================/// ///==========================================================================///
// RUN: %clang_cc1 -DCK5 -verify -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK5 --check-prefix CK5-64 // RUN: %clang_cc1 -DCK5 -verify -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK5 --check-prefix CK5-64
// RUN: %clang_cc1 -DCK5 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK5 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
▲ Show 20 Lines • Show All 169 Lines • Show Last 20 Lines

clang/test/OpenMP/target_data_map_codegen_hold.cpp

// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --check-globals --global-value-regex ".offload_maptypes.*" ".offload_sizes.*" --global-hex-value-regex ".offload_maptypes.*" // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --check-globals --prefix-filecheck-ir-name _ --global-value-regex ".offload_maptypes.*" ".offload_sizes.*" --global-hex-value-regex ".offload_maptypes.*"
// expected-no-diagnostics // expected-no-diagnostics
#ifndef HEADER #ifndef HEADER
#define HEADER #define HEADER
// powerpc64le-ibm-linux-gnu // powerpc64le-ibm-linux-gnu
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-extensions \ // RUN: %clang_cc1 -verify -fopenmp -fopenmp-extensions \
// RUN: -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ \ // RUN: -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ \
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
//. //.
// CHECK-PPC64LE: @.offload_sizes = private unnamed_addr constant [1 x i64] [i64 20] // CHECK-PPC64LE: @.offload_sizes = private unnamed_addr constant [1 x i64] [i64 20]
// CHECK-PPC64LE: @.offload_maptypes = private unnamed_addr constant [1 x i64] [i64 [[#0x2001]]] // CHECK-PPC64LE: @.offload_maptypes = private unnamed_addr constant [1 x i64] [i64 [[#0x2001]]]
// CHECK-PPC64LE: @.offload_sizes.1 = private unnamed_addr constant [1 x i64] [i64 20] // CHECK-PPC64LE: @.offload_sizes.1 = private unnamed_addr constant [1 x i64] [i64 20]
// CHECK-PPC64LE: @.offload_maptypes.2 = private unnamed_addr constant [1 x i64] [i64 [[#0x2405]]] // CHECK-PPC64LE: @.offload_maptypes.2 = private unnamed_addr constant [1 x i64] [i64 [[#0x2405]]]
// CHECK-PPC64LE: @.offload_sizes.3 = private unnamed_addr constant [1 x i64] [i64 4] // CHECK-PPC64LE: @.offload_sizes.3 = private unnamed_addr constant [1 x i64] [i64 4]
// CHECK-PPC64LE: @.offload_maptypes.4 = private unnamed_addr constant [1 x i64] [i64 [[#0x2003]]] // CHECK-PPC64LE: @.offload_maptypes.4 = private unnamed_addr constant [1 x i64] [i64 [[#0x2003]]]
// CHECK-PPC64LE: @.offload_maptypes.5 = private unnamed_addr constant [11 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002010]], i64 [[#0x2010]], i64 [[#0x2013]], i64 [[#0x3]], i64 [[#0x2000]], i64 [[#0x7000000002003]], i64 [[#0x7000000002010]], i64 [[#0x2010]], i64 [[#0x2013]]] // CHECK-PPC64LE: @.offload_sizes.5 = private unnamed_addr global [11 x i64] [i64 0, i64 4, i64 8, i64 8, i64 4, i64 4, i64 0, i64 4, i64 8, i64 8, i64 4]
// CHECK-PPC64LE: @.offload_maptypes.6 = private unnamed_addr constant [11 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002010]], i64 [[#0x2010]], i64 [[#0x2013]], i64 [[#0x3]], i64 [[#0x2000]], i64 [[#0x7000000002003]], i64 [[#0x7000000002010]], i64 [[#0x2010]], i64 [[#0x2013]]]
//. //.
// CHECK-I386: @.offload_sizes = private unnamed_addr constant [1 x i64] [i64 20] // CHECK-I386: @.offload_sizes = private unnamed_addr constant [1 x i64] [i64 20]
// CHECK-I386: @.offload_maptypes = private unnamed_addr constant [1 x i64] [i64 [[#0x2001]]] // CHECK-I386: @.offload_maptypes = private unnamed_addr constant [1 x i64] [i64 [[#0x2001]]]
// CHECK-I386: @.offload_sizes.1 = private unnamed_addr constant [1 x i64] [i64 20] // CHECK-I386: @.offload_sizes.1 = private unnamed_addr constant [1 x i64] [i64 20]
// CHECK-I386: @.offload_maptypes.2 = private unnamed_addr constant [1 x i64] [i64 [[#0x2405]]] // CHECK-I386: @.offload_maptypes.2 = private unnamed_addr constant [1 x i64] [i64 [[#0x2405]]]
// CHECK-I386: @.offload_sizes.3 = private unnamed_addr constant [1 x i64] [i64 4] // CHECK-I386: @.offload_sizes.3 = private unnamed_addr constant [1 x i64] [i64 4]
// CHECK-I386: @.offload_maptypes.4 = private unnamed_addr constant [1 x i64] [i64 [[#0x2003]]] // CHECK-I386: @.offload_maptypes.4 = private unnamed_addr constant [1 x i64] [i64 [[#0x2003]]]
// CHECK-I386: @.offload_maptypes.5 = private unnamed_addr constant [11 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002010]], i64 [[#0x2010]], i64 [[#0x2013]], i64 [[#0x3]], i64 [[#0x2000]], i64 [[#0x7000000002003]], i64 [[#0x7000000002010]], i64 [[#0x2010]], i64 [[#0x2013]]] // CHECK-I386: @.offload_sizes.5 = private unnamed_addr global [11 x i64] [i64 0, i64 4, i64 4, i64 4, i64 4, i64 4, i64 0, i64 4, i64 4, i64 4, i64 4]
// CHECK-I386: @.offload_maptypes.6 = private unnamed_addr constant [11 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002010]], i64 [[#0x2010]], i64 [[#0x2013]], i64 [[#0x3]], i64 [[#0x2000]], i64 [[#0x7000000002003]], i64 [[#0x7000000002010]], i64 [[#0x2010]], i64 [[#0x2013]]]
//. //.
// CHECK-PPC64LE-LABEL: @_Z3fooi( // CHECK-PPC64LE-LABEL: @_Z3fooi(
// CHECK-PPC64LE-NEXT: entry: // CHECK-PPC64LE-NEXT: entry:
// CHECK-PPC64LE-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4 // CHECK-PPC64LE-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
// CHECK-PPC64LE-NEXT: [[LB:%.*]] = alloca [5 x float], align 4 // CHECK-PPC64LE-NEXT: [[LB:%.*]] = alloca [5 x float], align 4
// CHECK-PPC64LE-NEXT: [[PS1:%.*]] = alloca %struct.S2*, align 8 // CHECK-PPC64LE-NEXT: [[PS1:%.*]] = alloca %struct.S2*, align 8
// CHECK-PPC64LE-NEXT: [[PS2:%.*]] = alloca %struct.S2*, align 8 // CHECK-PPC64LE-NEXT: [[PS2:%.*]] = alloca %struct.S2*, align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [1 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [1 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [1 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [1 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [1 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [1 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [11 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [11 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [11 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [11 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [11 x i8*], align 8 // CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [11 x i8*], align 8
// CHECK-PPC64LE-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [11 x i64], align 8
// CHECK-PPC64LE-NEXT: store i32 [[ARG:%.*]], i32* [[ARG_ADDR]], align 4 // CHECK-PPC64LE-NEXT: store i32 [[ARG:%.*]], i32* [[ARG_ADDR]], align 4
// CHECK-PPC64LE-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK-PPC64LE-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: [[TMP1:%.*]] = bitcast i8** [[TMP0]] to [5 x float]** // CHECK-PPC64LE-NEXT: [[TMP1:%.*]] = bitcast i8** [[TMP0]] to [5 x float]**
// CHECK-PPC64LE-NEXT: store [5 x float]* [[LB]], [5 x float]** [[TMP1]], align 8 // CHECK-PPC64LE-NEXT: store [5 x float]* [[LB]], [5 x float]** [[TMP1]], align 8
// CHECK-PPC64LE-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK-PPC64LE-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [5 x float]** // CHECK-PPC64LE-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [5 x float]**
// CHECK-PPC64LE-NEXT: store [5 x float]* [[LB]], [5 x float]** [[TMP3]], align 8 // CHECK-PPC64LE-NEXT: store [5 x float]* [[LB]], [5 x float]** [[TMP3]], align 8
// CHECK-PPC64LE-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK-PPC64LE-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
▲ Show 20 Lines • Show All 104 Lines • ▼ Show 20 Lines
// CHECK-PPC64LE-NEXT: [[TMP68:%.*]] = sub i64 [[TMP66]], [[TMP67]] // CHECK-PPC64LE-NEXT: [[TMP68:%.*]] = sub i64 [[TMP66]], [[TMP67]]
// CHECK-PPC64LE-NEXT: [[TMP69:%.*]] = sdiv exact i64 [[TMP68]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK-PPC64LE-NEXT: [[TMP69:%.*]] = sdiv exact i64 [[TMP68]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK-PPC64LE-NEXT: [[TMP70:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0 // CHECK-PPC64LE-NEXT: [[TMP70:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to %struct.S2** // CHECK-PPC64LE-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to %struct.S2**
// CHECK-PPC64LE-NEXT: store %struct.S2* [[TMP30]], %struct.S2** [[TMP71]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S2* [[TMP30]], %struct.S2** [[TMP71]], align 8
// CHECK-PPC64LE-NEXT: [[TMP72:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0 // CHECK-PPC64LE-NEXT: [[TMP72:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to %struct.S1** // CHECK-PPC64LE-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to %struct.S1**
// CHECK-PPC64LE-NEXT: store %struct.S1* [[S]], %struct.S1** [[TMP73]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S1* [[S]], %struct.S1** [[TMP73]], align 8
// CHECK-PPC64LE-NEXT: [[TMP74:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-PPC64LE-NEXT: store i64 [[TMP49]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_sizes.5, i32 0, i32 0), align 8
// CHECK-PPC64LE-NEXT: store i64 [[TMP49]], i64* [[TMP74]], align 8 // CHECK-PPC64LE-NEXT: [[TMP74:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 0
// CHECK-PPC64LE-NEXT: [[TMP75:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 0 // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP74]], align 8
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP75]], align 8 // CHECK-PPC64LE-NEXT: [[TMP75:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1
// CHECK-PPC64LE-NEXT: [[TMP76:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1 // CHECK-PPC64LE-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to %struct.S2**
// CHECK-PPC64LE-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to %struct.S2** // CHECK-PPC64LE-NEXT: store %struct.S2* [[TMP30]], %struct.S2** [[TMP76]], align 8
// CHECK-PPC64LE-NEXT: store %struct.S2* [[TMP30]], %struct.S2** [[TMP77]], align 8 // CHECK-PPC64LE-NEXT: [[TMP77:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1
// CHECK-PPC64LE-NEXT: [[TMP78:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1 // CHECK-PPC64LE-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to %struct.S1**
// CHECK-PPC64LE-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to %struct.S1** // CHECK-PPC64LE-NEXT: store %struct.S1* [[S]], %struct.S1** [[TMP78]], align 8
// CHECK-PPC64LE-NEXT: store %struct.S1* [[S]], %struct.S1** [[TMP79]], align 8 // CHECK-PPC64LE-NEXT: [[TMP79:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 1
// CHECK-PPC64LE-NEXT: [[TMP80:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP79]], align 8
// CHECK-PPC64LE-NEXT: store i64 4, i64* [[TMP80]], align 8 // CHECK-PPC64LE-NEXT: [[TMP80:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2
// CHECK-PPC64LE-NEXT: [[TMP81:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 1 // CHECK-PPC64LE-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to %struct.S2***
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP81]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S2** [[PS]], %struct.S2*** [[TMP81]], align 8
// CHECK-PPC64LE-NEXT: [[TMP82:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2 // CHECK-PPC64LE-NEXT: [[TMP82:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2
// CHECK-PPC64LE-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to %struct.S2*** // CHECK-PPC64LE-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to %struct.S2***
// CHECK-PPC64LE-NEXT: store %struct.S2** [[PS]], %struct.S2*** [[TMP83]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S2** [[PS10]], %struct.S2*** [[TMP83]], align 8
// CHECK-PPC64LE-NEXT: [[TMP84:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2 // CHECK-PPC64LE-NEXT: [[TMP84:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 2
// CHECK-PPC64LE-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to %struct.S2*** // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP84]], align 8
// CHECK-PPC64LE-NEXT: store %struct.S2** [[PS10]], %struct.S2*** [[TMP85]], align 8 // CHECK-PPC64LE-NEXT: [[TMP85:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3
// CHECK-PPC64LE-NEXT: [[TMP86:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK-PPC64LE-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to %struct.S2***
// CHECK-PPC64LE-NEXT: store i64 8, i64* [[TMP86]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S2** [[PS10]], %struct.S2*** [[TMP86]], align 8
// CHECK-PPC64LE-NEXT: [[TMP87:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 2 // CHECK-PPC64LE-NEXT: [[TMP87:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP87]], align 8 // CHECK-PPC64LE-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to %struct.S2***
// CHECK-PPC64LE-NEXT: [[TMP88:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3 // CHECK-PPC64LE-NEXT: store %struct.S2** [[PS13]], %struct.S2*** [[TMP88]], align 8
// CHECK-PPC64LE-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to %struct.S2*** // CHECK-PPC64LE-NEXT: [[TMP89:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 3
// CHECK-PPC64LE-NEXT: store %struct.S2** [[PS10]], %struct.S2*** [[TMP89]], align 8 // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP89]], align 8
// CHECK-PPC64LE-NEXT: [[TMP90:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3 // CHECK-PPC64LE-NEXT: [[TMP90:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4
// CHECK-PPC64LE-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to %struct.S2*** // CHECK-PPC64LE-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to %struct.S2***
// CHECK-PPC64LE-NEXT: store %struct.S2** [[PS13]], %struct.S2*** [[TMP91]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S2** [[PS13]], %struct.S2*** [[TMP91]], align 8
// CHECK-PPC64LE-NEXT: [[TMP92:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK-PPC64LE-NEXT: [[TMP92:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4
// CHECK-PPC64LE-NEXT: store i64 8, i64* [[TMP92]], align 8 // CHECK-PPC64LE-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to %struct.S1**
// CHECK-PPC64LE-NEXT: [[TMP93:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 3 // CHECK-PPC64LE-NEXT: store %struct.S1* [[S17]], %struct.S1** [[TMP93]], align 8
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP93]], align 8 // CHECK-PPC64LE-NEXT: [[TMP94:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 4
// CHECK-PPC64LE-NEXT: [[TMP94:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4 // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP94]], align 8
// CHECK-PPC64LE-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to %struct.S2*** // CHECK-PPC64LE-NEXT: [[TMP95:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5
// CHECK-PPC64LE-NEXT: store %struct.S2** [[PS13]], %struct.S2*** [[TMP95]], align 8 // CHECK-PPC64LE-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i32**
// CHECK-PPC64LE-NEXT: [[TMP96:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4 // CHECK-PPC64LE-NEXT: store i32* [[ARG_ADDR]], i32** [[TMP96]], align 8
// CHECK-PPC64LE-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to %struct.S1** // CHECK-PPC64LE-NEXT: [[TMP97:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5
// CHECK-PPC64LE-NEXT: store %struct.S1* [[S17]], %struct.S1** [[TMP97]], align 8 // CHECK-PPC64LE-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32**
// CHECK-PPC64LE-NEXT: [[TMP98:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK-PPC64LE-NEXT: store i32* [[ARG_ADDR]], i32** [[TMP98]], align 8
// CHECK-PPC64LE-NEXT: store i64 4, i64* [[TMP98]], align 8 // CHECK-PPC64LE-NEXT: [[TMP99:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 5
// CHECK-PPC64LE-NEXT: [[TMP99:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 4
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP99]], align 8 // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP99]], align 8
// CHECK-PPC64LE-NEXT: [[TMP100:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5 // CHECK-PPC64LE-NEXT: [[TMP100:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6
// CHECK-PPC64LE-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i32** // CHECK-PPC64LE-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to %struct.S2**
// CHECK-PPC64LE-NEXT: store i32* [[ARG_ADDR]], i32** [[TMP101]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S2* [[TMP50]], %struct.S2** [[TMP101]], align 8
// CHECK-PPC64LE-NEXT: [[TMP102:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5 // CHECK-PPC64LE-NEXT: [[TMP102:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6
// CHECK-PPC64LE-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to i32** // CHECK-PPC64LE-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to %struct.S1**
// CHECK-PPC64LE-NEXT: store i32* [[ARG_ADDR]], i32** [[TMP103]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S1* [[S18]], %struct.S1** [[TMP103]], align 8
// CHECK-PPC64LE-NEXT: [[TMP104:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK-PPC64LE-NEXT: store i64 [[TMP69]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_sizes.5, i32 0, i32 6), align 8
// CHECK-PPC64LE-NEXT: store i64 4, i64* [[TMP104]], align 8 // CHECK-PPC64LE-NEXT: [[TMP104:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 6
// CHECK-PPC64LE-NEXT: [[TMP105:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 5 // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP104]], align 8
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP105]], align 8 // CHECK-PPC64LE-NEXT: [[TMP105:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7
// CHECK-PPC64LE-NEXT: [[TMP106:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6 // CHECK-PPC64LE-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to %struct.S2**
// CHECK-PPC64LE-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to %struct.S2** // CHECK-PPC64LE-NEXT: store %struct.S2* [[TMP50]], %struct.S2** [[TMP106]], align 8
// CHECK-PPC64LE-NEXT: store %struct.S2* [[TMP50]], %struct.S2** [[TMP107]], align 8 // CHECK-PPC64LE-NEXT: [[TMP107:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7
// CHECK-PPC64LE-NEXT: [[TMP108:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6 // CHECK-PPC64LE-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to %struct.S1**
// CHECK-PPC64LE-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to %struct.S1** // CHECK-PPC64LE-NEXT: store %struct.S1* [[S18]], %struct.S1** [[TMP108]], align 8
// CHECK-PPC64LE-NEXT: store %struct.S1* [[S18]], %struct.S1** [[TMP109]], align 8 // CHECK-PPC64LE-NEXT: [[TMP109:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 7
// CHECK-PPC64LE-NEXT: [[TMP110:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP109]], align 8
// CHECK-PPC64LE-NEXT: store i64 [[TMP69]], i64* [[TMP110]], align 8 // CHECK-PPC64LE-NEXT: [[TMP110:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8
// CHECK-PPC64LE-NEXT: [[TMP111:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 6 // CHECK-PPC64LE-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to %struct.S2***
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP111]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S2** [[PS19]], %struct.S2*** [[TMP111]], align 8
// CHECK-PPC64LE-NEXT: [[TMP112:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7 // CHECK-PPC64LE-NEXT: [[TMP112:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8
// CHECK-PPC64LE-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to %struct.S2** // CHECK-PPC64LE-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to %struct.S2***
// CHECK-PPC64LE-NEXT: store %struct.S2* [[TMP50]], %struct.S2** [[TMP113]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S2** [[PS21]], %struct.S2*** [[TMP113]], align 8
// CHECK-PPC64LE-NEXT: [[TMP114:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7 // CHECK-PPC64LE-NEXT: [[TMP114:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 8
// CHECK-PPC64LE-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to %struct.S1** // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP114]], align 8
// CHECK-PPC64LE-NEXT: store %struct.S1* [[S18]], %struct.S1** [[TMP115]], align 8 // CHECK-PPC64LE-NEXT: [[TMP115:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9
// CHECK-PPC64LE-NEXT: [[TMP116:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK-PPC64LE-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to %struct.S2***
// CHECK-PPC64LE-NEXT: store i64 4, i64* [[TMP116]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S2** [[PS21]], %struct.S2*** [[TMP116]], align 8
// CHECK-PPC64LE-NEXT: [[TMP117:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 7 // CHECK-PPC64LE-NEXT: [[TMP117:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP117]], align 8 // CHECK-PPC64LE-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to %struct.S2***
// CHECK-PPC64LE-NEXT: [[TMP118:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8 // CHECK-PPC64LE-NEXT: store %struct.S2** [[PS24]], %struct.S2*** [[TMP118]], align 8
// CHECK-PPC64LE-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to %struct.S2*** // CHECK-PPC64LE-NEXT: [[TMP119:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 9
// CHECK-PPC64LE-NEXT: store %struct.S2** [[PS19]], %struct.S2*** [[TMP119]], align 8 // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP119]], align 8
// CHECK-PPC64LE-NEXT: [[TMP120:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8 // CHECK-PPC64LE-NEXT: [[TMP120:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 10
// CHECK-PPC64LE-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to %struct.S2*** // CHECK-PPC64LE-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to %struct.S2***
// CHECK-PPC64LE-NEXT: store %struct.S2** [[PS21]], %struct.S2*** [[TMP121]], align 8 // CHECK-PPC64LE-NEXT: store %struct.S2** [[PS24]], %struct.S2*** [[TMP121]], align 8
// CHECK-PPC64LE-NEXT: [[TMP122:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK-PPC64LE-NEXT: [[TMP122:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 10
// CHECK-PPC64LE-NEXT: store i64 8, i64* [[TMP122]], align 8 // CHECK-PPC64LE-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to %struct.S1**
// CHECK-PPC64LE-NEXT: [[TMP123:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 8 // CHECK-PPC64LE-NEXT: store %struct.S1* [[S28]], %struct.S1** [[TMP123]], align 8
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP123]], align 8 // CHECK-PPC64LE-NEXT: [[TMP124:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 10
// CHECK-PPC64LE-NEXT: [[TMP124:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9 // CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP124]], align 8
// CHECK-PPC64LE-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to %struct.S2*** // CHECK-PPC64LE-NEXT: [[TMP125:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: store %struct.S2** [[PS21]], %struct.S2*** [[TMP125]], align 8 // CHECK-PPC64LE-NEXT: [[TMP126:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: [[TMP126:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9 // CHECK-PPC64LE-NEXT: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i32 11, i8** [[TMP125]], i8** [[TMP126]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null)
// CHECK-PPC64LE-NEXT: [[TMP127:%.*]] = bitcast i8** [[TMP126]] to %struct.S2*** // CHECK-PPC64LE-NEXT: [[TMP127:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
// CHECK-PPC64LE-NEXT: store %struct.S2** [[PS24]], %struct.S2*** [[TMP127]], align 8 // CHECK-PPC64LE-NEXT: [[INC32:%.*]] = add nsw i32 [[TMP127]], 1
// CHECK-PPC64LE-NEXT: [[TMP128:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK-PPC64LE-NEXT: store i64 8, i64* [[TMP128]], align 8
// CHECK-PPC64LE-NEXT: [[TMP129:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 9
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP129]], align 8
// CHECK-PPC64LE-NEXT: [[TMP130:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 10
// CHECK-PPC64LE-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to %struct.S2***
// CHECK-PPC64LE-NEXT: store %struct.S2** [[PS24]], %struct.S2*** [[TMP131]], align 8
// CHECK-PPC64LE-NEXT: [[TMP132:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 10
// CHECK-PPC64LE-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to %struct.S1**
// CHECK-PPC64LE-NEXT: store %struct.S1* [[S28]], %struct.S1** [[TMP133]], align 8
// CHECK-PPC64LE-NEXT: [[TMP134:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 10
// CHECK-PPC64LE-NEXT: store i64 4, i64* [[TMP134]], align 8
// CHECK-PPC64LE-NEXT: [[TMP135:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 10
// CHECK-PPC64LE-NEXT: store i8* null, i8** [[TMP135]], align 8
// CHECK-PPC64LE-NEXT: [[TMP136:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: [[TMP137:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: [[TMP138:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i32 11, i8** [[TMP136]], i8** [[TMP137]], i64* [[TMP138]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null)
// CHECK-PPC64LE-NEXT: [[TMP139:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
// CHECK-PPC64LE-NEXT: [[INC32:%.*]] = add nsw i32 [[TMP139]], 1
// CHECK-PPC64LE-NEXT: store i32 [[INC32]], i32* [[ARG_ADDR]], align 4 // CHECK-PPC64LE-NEXT: store i32 [[INC32]], i32* [[ARG_ADDR]], align 4
// CHECK-PPC64LE-NEXT: [[TMP140:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0 // CHECK-PPC64LE-NEXT: [[TMP128:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: [[TMP141:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0 // CHECK-PPC64LE-NEXT: [[TMP129:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK-PPC64LE-NEXT: [[TMP142:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-PPC64LE-NEXT: call void @__tgt_target_data_end_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i32 11, i8** [[TMP128]], i8** [[TMP129]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null)
// CHECK-PPC64LE-NEXT: call void @__tgt_target_data_end_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i32 11, i8** [[TMP140]], i8** [[TMP141]], i64* [[TMP142]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null)
// CHECK-PPC64LE-NEXT: ret void // CHECK-PPC64LE-NEXT: ret void
// //
// CHECK-I386-LABEL: @_Z3fooi( // CHECK-I386-LABEL: @_Z3fooi(
// CHECK-I386-NEXT: entry: // CHECK-I386-NEXT: entry:
// CHECK-I386-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4 // CHECK-I386-NEXT: [[ARG_ADDR:%.*]] = alloca i32, align 4
// CHECK-I386-NEXT: [[LB:%.*]] = alloca [5 x float], align 4 // CHECK-I386-NEXT: [[LB:%.*]] = alloca [5 x float], align 4
// CHECK-I386-NEXT: [[PS1:%.*]] = alloca %struct.S2*, align 4 // CHECK-I386-NEXT: [[PS1:%.*]] = alloca %struct.S2*, align 4
// CHECK-I386-NEXT: [[PS2:%.*]] = alloca %struct.S2*, align 4 // CHECK-I386-NEXT: [[PS2:%.*]] = alloca %struct.S2*, align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [1 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [1 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [1 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [1 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [1 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [1 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [11 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [11 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [11 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [11 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [11 x i8*], align 4 // CHECK-I386-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [11 x i8*], align 4
// CHECK-I386-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [11 x i64], align 4
// CHECK-I386-NEXT: store i32 [[ARG:%.*]], i32* [[ARG_ADDR]], align 4 // CHECK-I386-NEXT: store i32 [[ARG:%.*]], i32* [[ARG_ADDR]], align 4
// CHECK-I386-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK-I386-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-I386-NEXT: [[TMP1:%.*]] = bitcast i8** [[TMP0]] to [5 x float]** // CHECK-I386-NEXT: [[TMP1:%.*]] = bitcast i8** [[TMP0]] to [5 x float]**
// CHECK-I386-NEXT: store [5 x float]* [[LB]], [5 x float]** [[TMP1]], align 4 // CHECK-I386-NEXT: store [5 x float]* [[LB]], [5 x float]** [[TMP1]], align 4
// CHECK-I386-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK-I386-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-I386-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [5 x float]** // CHECK-I386-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [5 x float]**
// CHECK-I386-NEXT: store [5 x float]* [[LB]], [5 x float]** [[TMP3]], align 4 // CHECK-I386-NEXT: store [5 x float]* [[LB]], [5 x float]** [[TMP3]], align 4
// CHECK-I386-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK-I386-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
▲ Show 20 Lines • Show All 104 Lines • ▼ Show 20 Lines
// CHECK-I386-NEXT: [[TMP68:%.*]] = sub i64 [[TMP66]], [[TMP67]] // CHECK-I386-NEXT: [[TMP68:%.*]] = sub i64 [[TMP66]], [[TMP67]]
// CHECK-I386-NEXT: [[TMP69:%.*]] = sdiv exact i64 [[TMP68]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK-I386-NEXT: [[TMP69:%.*]] = sdiv exact i64 [[TMP68]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK-I386-NEXT: [[TMP70:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0 // CHECK-I386-NEXT: [[TMP70:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK-I386-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to %struct.S2** // CHECK-I386-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to %struct.S2**
// CHECK-I386-NEXT: store %struct.S2* [[TMP30]], %struct.S2** [[TMP71]], align 4 // CHECK-I386-NEXT: store %struct.S2* [[TMP30]], %struct.S2** [[TMP71]], align 4
// CHECK-I386-NEXT: [[TMP72:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0 // CHECK-I386-NEXT: [[TMP72:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK-I386-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to %struct.S1** // CHECK-I386-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to %struct.S1**
// CHECK-I386-NEXT: store %struct.S1* [[S]], %struct.S1** [[TMP73]], align 4 // CHECK-I386-NEXT: store %struct.S1* [[S]], %struct.S1** [[TMP73]], align 4
// CHECK-I386-NEXT: [[TMP74:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-I386-NEXT: store i64 [[TMP49]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_sizes.5, i32 0, i32 0), align 4
// CHECK-I386-NEXT: store i64 [[TMP49]], i64* [[TMP74]], align 4 // CHECK-I386-NEXT: [[TMP74:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 0
// CHECK-I386-NEXT: [[TMP75:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 0 // CHECK-I386-NEXT: store i8* null, i8** [[TMP74]], align 4
// CHECK-I386-NEXT: store i8* null, i8** [[TMP75]], align 4 // CHECK-I386-NEXT: [[TMP75:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1
// CHECK-I386-NEXT: [[TMP76:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1 // CHECK-I386-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to %struct.S2**
// CHECK-I386-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to %struct.S2** // CHECK-I386-NEXT: store %struct.S2* [[TMP30]], %struct.S2** [[TMP76]], align 4
// CHECK-I386-NEXT: store %struct.S2* [[TMP30]], %struct.S2** [[TMP77]], align 4 // CHECK-I386-NEXT: [[TMP77:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1
// CHECK-I386-NEXT: [[TMP78:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1 // CHECK-I386-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to %struct.S1**
// CHECK-I386-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to %struct.S1** // CHECK-I386-NEXT: store %struct.S1* [[S]], %struct.S1** [[TMP78]], align 4
// CHECK-I386-NEXT: store %struct.S1* [[S]], %struct.S1** [[TMP79]], align 4 // CHECK-I386-NEXT: [[TMP79:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 1
// CHECK-I386-NEXT: [[TMP80:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK-I386-NEXT: store i8* null, i8** [[TMP79]], align 4
// CHECK-I386-NEXT: store i64 4, i64* [[TMP80]], align 4 // CHECK-I386-NEXT: [[TMP80:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2
// CHECK-I386-NEXT: [[TMP81:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 1 // CHECK-I386-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to %struct.S2***
// CHECK-I386-NEXT: store i8* null, i8** [[TMP81]], align 4 // CHECK-I386-NEXT: store %struct.S2** [[PS]], %struct.S2*** [[TMP81]], align 4
// CHECK-I386-NEXT: [[TMP82:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2 // CHECK-I386-NEXT: [[TMP82:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2
// CHECK-I386-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to %struct.S2*** // CHECK-I386-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to %struct.S2***
// CHECK-I386-NEXT: store %struct.S2** [[PS]], %struct.S2*** [[TMP83]], align 4 // CHECK-I386-NEXT: store %struct.S2** [[PS10]], %struct.S2*** [[TMP83]], align 4
// CHECK-I386-NEXT: [[TMP84:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2 // CHECK-I386-NEXT: [[TMP84:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 2
// CHECK-I386-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to %struct.S2*** // CHECK-I386-NEXT: store i8* null, i8** [[TMP84]], align 4
// CHECK-I386-NEXT: store %struct.S2** [[PS10]], %struct.S2*** [[TMP85]], align 4 // CHECK-I386-NEXT: [[TMP85:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3
// CHECK-I386-NEXT: [[TMP86:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK-I386-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to %struct.S2***
// CHECK-I386-NEXT: store i64 4, i64* [[TMP86]], align 4 // CHECK-I386-NEXT: store %struct.S2** [[PS10]], %struct.S2*** [[TMP86]], align 4
// CHECK-I386-NEXT: [[TMP87:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 2 // CHECK-I386-NEXT: [[TMP87:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3
// CHECK-I386-NEXT: store i8* null, i8** [[TMP87]], align 4 // CHECK-I386-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to %struct.S2***
// CHECK-I386-NEXT: [[TMP88:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3 // CHECK-I386-NEXT: store %struct.S2** [[PS13]], %struct.S2*** [[TMP88]], align 4
// CHECK-I386-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to %struct.S2*** // CHECK-I386-NEXT: [[TMP89:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 3
// CHECK-I386-NEXT: store %struct.S2** [[PS10]], %struct.S2*** [[TMP89]], align 4 // CHECK-I386-NEXT: store i8* null, i8** [[TMP89]], align 4
// CHECK-I386-NEXT: [[TMP90:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3 // CHECK-I386-NEXT: [[TMP90:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4
// CHECK-I386-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to %struct.S2*** // CHECK-I386-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to %struct.S2***
// CHECK-I386-NEXT: store %struct.S2** [[PS13]], %struct.S2*** [[TMP91]], align 4 // CHECK-I386-NEXT: store %struct.S2** [[PS13]], %struct.S2*** [[TMP91]], align 4
// CHECK-I386-NEXT: [[TMP92:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK-I386-NEXT: [[TMP92:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4
// CHECK-I386-NEXT: store i64 4, i64* [[TMP92]], align 4 // CHECK-I386-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to %struct.S1**
// CHECK-I386-NEXT: [[TMP93:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 3 // CHECK-I386-NEXT: store %struct.S1* [[S17]], %struct.S1** [[TMP93]], align 4
// CHECK-I386-NEXT: store i8* null, i8** [[TMP93]], align 4 // CHECK-I386-NEXT: [[TMP94:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 4
// CHECK-I386-NEXT: [[TMP94:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4 // CHECK-I386-NEXT: store i8* null, i8** [[TMP94]], align 4
// CHECK-I386-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to %struct.S2*** // CHECK-I386-NEXT: [[TMP95:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5
// CHECK-I386-NEXT: store %struct.S2** [[PS13]], %struct.S2*** [[TMP95]], align 4 // CHECK-I386-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i32**
// CHECK-I386-NEXT: [[TMP96:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4 // CHECK-I386-NEXT: store i32* [[ARG_ADDR]], i32** [[TMP96]], align 4
// CHECK-I386-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to %struct.S1** // CHECK-I386-NEXT: [[TMP97:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5
// CHECK-I386-NEXT: store %struct.S1* [[S17]], %struct.S1** [[TMP97]], align 4 // CHECK-I386-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32**
// CHECK-I386-NEXT: [[TMP98:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK-I386-NEXT: store i32* [[ARG_ADDR]], i32** [[TMP98]], align 4
// CHECK-I386-NEXT: store i64 4, i64* [[TMP98]], align 4 // CHECK-I386-NEXT: [[TMP99:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 5
// CHECK-I386-NEXT: [[TMP99:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 4
// CHECK-I386-NEXT: store i8* null, i8** [[TMP99]], align 4 // CHECK-I386-NEXT: store i8* null, i8** [[TMP99]], align 4
// CHECK-I386-NEXT: [[TMP100:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5 // CHECK-I386-NEXT: [[TMP100:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6
// CHECK-I386-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i32** // CHECK-I386-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to %struct.S2**
// CHECK-I386-NEXT: store i32* [[ARG_ADDR]], i32** [[TMP101]], align 4 // CHECK-I386-NEXT: store %struct.S2* [[TMP50]], %struct.S2** [[TMP101]], align 4
// CHECK-I386-NEXT: [[TMP102:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5 // CHECK-I386-NEXT: [[TMP102:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6
// CHECK-I386-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to i32** // CHECK-I386-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to %struct.S1**
// CHECK-I386-NEXT: store i32* [[ARG_ADDR]], i32** [[TMP103]], align 4 // CHECK-I386-NEXT: store %struct.S1* [[S18]], %struct.S1** [[TMP103]], align 4
// CHECK-I386-NEXT: [[TMP104:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK-I386-NEXT: store i64 [[TMP69]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_sizes.5, i32 0, i32 6), align 4
// CHECK-I386-NEXT: store i64 4, i64* [[TMP104]], align 4 // CHECK-I386-NEXT: [[TMP104:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 6
// CHECK-I386-NEXT: [[TMP105:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 5 // CHECK-I386-NEXT: store i8* null, i8** [[TMP104]], align 4
// CHECK-I386-NEXT: store i8* null, i8** [[TMP105]], align 4 // CHECK-I386-NEXT: [[TMP105:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7
// CHECK-I386-NEXT: [[TMP106:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6 // CHECK-I386-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to %struct.S2**
// CHECK-I386-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to %struct.S2** // CHECK-I386-NEXT: store %struct.S2* [[TMP50]], %struct.S2** [[TMP106]], align 4
// CHECK-I386-NEXT: store %struct.S2* [[TMP50]], %struct.S2** [[TMP107]], align 4 // CHECK-I386-NEXT: [[TMP107:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7
// CHECK-I386-NEXT: [[TMP108:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6 // CHECK-I386-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to %struct.S1**
// CHECK-I386-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to %struct.S1** // CHECK-I386-NEXT: store %struct.S1* [[S18]], %struct.S1** [[TMP108]], align 4
// CHECK-I386-NEXT: store %struct.S1* [[S18]], %struct.S1** [[TMP109]], align 4 // CHECK-I386-NEXT: [[TMP109:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 7
// CHECK-I386-NEXT: [[TMP110:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK-I386-NEXT: store i8* null, i8** [[TMP109]], align 4
// CHECK-I386-NEXT: store i64 [[TMP69]], i64* [[TMP110]], align 4 // CHECK-I386-NEXT: [[TMP110:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8
// CHECK-I386-NEXT: [[TMP111:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 6 // CHECK-I386-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to %struct.S2***
// CHECK-I386-NEXT: store i8* null, i8** [[TMP111]], align 4 // CHECK-I386-NEXT: store %struct.S2** [[PS19]], %struct.S2*** [[TMP111]], align 4
// CHECK-I386-NEXT: [[TMP112:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7 // CHECK-I386-NEXT: [[TMP112:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8
// CHECK-I386-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to %struct.S2** // CHECK-I386-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to %struct.S2***
// CHECK-I386-NEXT: store %struct.S2* [[TMP50]], %struct.S2** [[TMP113]], align 4 // CHECK-I386-NEXT: store %struct.S2** [[PS21]], %struct.S2*** [[TMP113]], align 4
// CHECK-I386-NEXT: [[TMP114:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7 // CHECK-I386-NEXT: [[TMP114:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 8
// CHECK-I386-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to %struct.S1** // CHECK-I386-NEXT: store i8* null, i8** [[TMP114]], align 4
// CHECK-I386-NEXT: store %struct.S1* [[S18]], %struct.S1** [[TMP115]], align 4 // CHECK-I386-NEXT: [[TMP115:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9
// CHECK-I386-NEXT: [[TMP116:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK-I386-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to %struct.S2***
// CHECK-I386-NEXT: store i64 4, i64* [[TMP116]], align 4 // CHECK-I386-NEXT: store %struct.S2** [[PS21]], %struct.S2*** [[TMP116]], align 4
// CHECK-I386-NEXT: [[TMP117:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 7 // CHECK-I386-NEXT: [[TMP117:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9
// CHECK-I386-NEXT: store i8* null, i8** [[TMP117]], align 4 // CHECK-I386-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to %struct.S2***
// CHECK-I386-NEXT: [[TMP118:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8 // CHECK-I386-NEXT: store %struct.S2** [[PS24]], %struct.S2*** [[TMP118]], align 4
// CHECK-I386-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to %struct.S2*** // CHECK-I386-NEXT: [[TMP119:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 9
// CHECK-I386-NEXT: store %struct.S2** [[PS19]], %struct.S2*** [[TMP119]], align 4 // CHECK-I386-NEXT: store i8* null, i8** [[TMP119]], align 4
// CHECK-I386-NEXT: [[TMP120:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8 // CHECK-I386-NEXT: [[TMP120:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 10
// CHECK-I386-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to %struct.S2*** // CHECK-I386-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to %struct.S2***
// CHECK-I386-NEXT: store %struct.S2** [[PS21]], %struct.S2*** [[TMP121]], align 4 // CHECK-I386-NEXT: store %struct.S2** [[PS24]], %struct.S2*** [[TMP121]], align 4
// CHECK-I386-NEXT: [[TMP122:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK-I386-NEXT: [[TMP122:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 10
// CHECK-I386-NEXT: store i64 4, i64* [[TMP122]], align 4 // CHECK-I386-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to %struct.S1**
// CHECK-I386-NEXT: [[TMP123:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 8 // CHECK-I386-NEXT: store %struct.S1* [[S28]], %struct.S1** [[TMP123]], align 4
// CHECK-I386-NEXT: store i8* null, i8** [[TMP123]], align 4 // CHECK-I386-NEXT: [[TMP124:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 10
// CHECK-I386-NEXT: [[TMP124:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9 // CHECK-I386-NEXT: store i8* null, i8** [[TMP124]], align 4
// CHECK-I386-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to %struct.S2*** // CHECK-I386-NEXT: [[TMP125:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK-I386-NEXT: store %struct.S2** [[PS21]], %struct.S2*** [[TMP125]], align 4 // CHECK-I386-NEXT: [[TMP126:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK-I386-NEXT: [[TMP126:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9 // CHECK-I386-NEXT: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i32 11, i8** [[TMP125]], i8** [[TMP126]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null)
// CHECK-I386-NEXT: [[TMP127:%.*]] = bitcast i8** [[TMP126]] to %struct.S2*** // CHECK-I386-NEXT: [[TMP127:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
// CHECK-I386-NEXT: store %struct.S2** [[PS24]], %struct.S2*** [[TMP127]], align 4 // CHECK-I386-NEXT: [[INC32:%.*]] = add nsw i32 [[TMP127]], 1
// CHECK-I386-NEXT: [[TMP128:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK-I386-NEXT: store i64 4, i64* [[TMP128]], align 4
// CHECK-I386-NEXT: [[TMP129:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 9
// CHECK-I386-NEXT: store i8* null, i8** [[TMP129]], align 4
// CHECK-I386-NEXT: [[TMP130:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 10
// CHECK-I386-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to %struct.S2***
// CHECK-I386-NEXT: store %struct.S2** [[PS24]], %struct.S2*** [[TMP131]], align 4
// CHECK-I386-NEXT: [[TMP132:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 10
// CHECK-I386-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to %struct.S1**
// CHECK-I386-NEXT: store %struct.S1* [[S28]], %struct.S1** [[TMP133]], align 4
// CHECK-I386-NEXT: [[TMP134:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 10
// CHECK-I386-NEXT: store i64 4, i64* [[TMP134]], align 4
// CHECK-I386-NEXT: [[TMP135:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i32 0, i32 10
// CHECK-I386-NEXT: store i8* null, i8** [[TMP135]], align 4
// CHECK-I386-NEXT: [[TMP136:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK-I386-NEXT: [[TMP137:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK-I386-NEXT: [[TMP138:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK-I386-NEXT: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i32 11, i8** [[TMP136]], i8** [[TMP137]], i64* [[TMP138]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null)
// CHECK-I386-NEXT: [[TMP139:%.*]] = load i32, i32* [[ARG_ADDR]], align 4
// CHECK-I386-NEXT: [[INC32:%.*]] = add nsw i32 [[TMP139]], 1
// CHECK-I386-NEXT: store i32 [[INC32]], i32* [[ARG_ADDR]], align 4 // CHECK-I386-NEXT: store i32 [[INC32]], i32* [[ARG_ADDR]], align 4
// CHECK-I386-NEXT: [[TMP140:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0 // CHECK-I386-NEXT: [[TMP128:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK-I386-NEXT: [[TMP141:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0 // CHECK-I386-NEXT: [[TMP129:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK-I386-NEXT: [[TMP142:%.*]] = getelementptr inbounds [11 x i64], [11 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-I386-NEXT: call void @__tgt_target_data_end_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i32 11, i8** [[TMP128]], i8** [[TMP129]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null)
// CHECK-I386-NEXT: call void @__tgt_target_data_end_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i32 11, i8** [[TMP140]], i8** [[TMP141]], i64* [[TMP142]], i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null)
// CHECK-I386-NEXT: ret void // CHECK-I386-NEXT: ret void
// //
void foo(int arg) { void foo(int arg) {
float lb[5]; float lb[5];
S2 *ps1; S2 *ps1;
S2 *ps2; S2 *ps2;
#pragma omp target data map(ompx_hold, to: lb) #pragma omp target data map(ompx_hold, to: lb)
Show All 19 Lines

clang/test/OpenMP/target_data_map_pointer_array_subscript_codegen.cpp

Show All 29 Linestypedef struct {
int *arr; int *arr;
} MyObject; } MyObject;
MyObject *objects; MyObject *objects;
#pragma omp end declare target #pragma omp end declare target
// CHECK-DAG: [[SIZES0:@.+]] = private unnamed_addr constant [1 x i64] [i64 {{8|4}}] // CHECK-DAG: [[SIZES0:@.+]] = private unnamed_addr constant [1 x i64] [i64 {{8|4}}]
// CHECK-DAG: [[MAPS0:@.+]] = private unnamed_addr constant [1 x i64] [i64 17] // CHECK-DAG: [[MAPS0:@.+]] = private unnamed_addr constant [1 x i64] [i64 17]
// CHECK-DAG: [[SIZES1:@.+]] = private unnamed_addr global [2 x i64] [i64 0, i64 4]
// CHECK-DAG: [[MAPS1:@.+]] = private unnamed_addr constant [2 x i64] [i64 0, i64 281474976710673] // CHECK-DAG: [[MAPS1:@.+]] = private unnamed_addr constant [2 x i64] [i64 0, i64 281474976710673]
// CHECK: @main // CHECK: @main
int main(void) { int main(void) {
// CHECK: [[BPTR0:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* %{{.+}}, i32 0, i32 0 // CHECK: [[BPTR0:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* %{{.+}}, i32 0, i32 0
// CHECK: [[BPTRC0:%.+]] = bitcast i8** [[BPTR0]] to %struct.MyObject*** // CHECK: [[BPTRC0:%.+]] = bitcast i8** [[BPTR0]] to %struct.MyObject***
// CHECK: store %struct.MyObject** @objects, %struct.MyObject*** [[BPTRC0]], // CHECK: store %struct.MyObject** @objects, %struct.MyObject*** [[BPTRC0]],
// CHECK: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 1, i8** %{{.+}}, i8** %{{.+}}, i64* getelementptr inbounds ([1 x i64], [1 x i64]* [[SIZES0]], i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* [[MAPS0]], i32 0, i32 0), i8** null, i8** null) // CHECK: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 1, i8** %{{.+}}, i8** %{{.+}}, i64* getelementptr inbounds ([1 x i64], [1 x i64]* [[SIZES0]], i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* [[MAPS0]], i32 0, i32 0), i8** null, i8** null)
#pragma omp target enter data map(to : objects [1:1]) #pragma omp target enter data map(to : objects [1:1])
// CHECK: [[OBJ:%.+]] = load %struct.MyObject*, %struct.MyObject** @objects, // CHECK: [[OBJ:%.+]] = load %struct.MyObject*, %struct.MyObject** @objects,
// CHECK: [[BPTR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* %{{.+}}, i32 0, i32 0 // CHECK: [[BPTR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* %{{.+}}, i32 0, i32 0
// CHECK: [[BPTRC0:%.+]] = bitcast i8** [[BPTR0]] to %struct.MyObject** // CHECK: [[BPTRC0:%.+]] = bitcast i8** [[BPTR0]] to %struct.MyObject**
// CHECK: store %struct.MyObject* [[OBJ]], %struct.MyObject** [[BPTRC0]], // CHECK: store %struct.MyObject* [[OBJ]], %struct.MyObject** [[BPTRC0]],
// CHECK: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** %{{.+}}, i8** %{{.+}}, i64* %{{.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[MAPS1]], i32 0, i32 0), i8** null, i8** null) // CHECK: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** %{{.+}}, i8** %{{.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZES1]], i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[MAPS1]], i32 0, i32 0), i8** null, i8** null)
#pragma omp target enter data map(to : objects[1].arr [0:1]) #pragma omp target enter data map(to : objects[1].arr [0:1])
return 0; return 0;
} }
#endif #endif

clang/test/OpenMP/target_data_use_device_addr_codegen.cpp

// RUN: %clang_cc1 -DCK1 -verify -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s // RUN: %clang_cc1 -DCK1 -verify -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s
// RUN: %clang_cc1 -DCK1 -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK1 -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s // RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
// RUN: %clang_cc1 -DCK1 -verify -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -DCK1 -verify -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -DCK1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}} // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
// expected-no-diagnostics // expected-no-diagnostics
#ifndef HEADER #ifndef HEADER
#define HEADER #define HEADER
// CHECK-DAG: [[SIZES1:@.+]] = private unnamed_addr constant [5 x i64] zeroinitializer // CHECK-DAG: [[SIZES1:@.+]] = private unnamed_addr constant [5 x i64] zeroinitializer
// 64 = 0x40 = OMP_MAP_RETURN_PARAM // 64 = 0x40 = OMP_MAP_RETURN_PARAM
// CHECK-DAG: [[MAPTYPES1:@.+]] = private unnamed_addr constant [5 x i64] [i64 64, i64 64, i64 64, i64 64, i64 64] // CHECK-DAG: [[MAPTYPES1:@.+]] = private unnamed_addr constant [5 x i64] [i64 64, i64 64, i64 64, i64 64, i64 64]
// CHECK-DAG: [[SIZES2:@.+]] = private unnamed_addr global [5 x i64] zeroinitializer
// 0 = OMP_MAP_NONE // 0 = OMP_MAP_NONE
// 281474976710720 = 0x1000000000040 = OMP_MAP_MEMBER_OF | OMP_MAP_RETURN_PARAM // 281474976710720 = 0x1000000000040 = OMP_MAP_MEMBER_OF | OMP_MAP_RETURN_PARAM
// CHECK-DAG: [[MAPTYPES2:@.+]] = private unnamed_addr constant [5 x i64] [i64 0, i64 281474976710720, i64 281474976710720, i64 281474976710720, i64 281474976710720] // CHECK-DAG: [[MAPTYPES2:@.+]] = private unnamed_addr constant [5 x i64] [i64 0, i64 281474976710720, i64 281474976710720, i64 281474976710720, i64 281474976710720]
struct S { struct S {
int a = 0; int a = 0;
int *ptr = &a; int *ptr = &a;
int &ref = a; int &ref = a;
int arr[4]; int arr[4];
▲ Show 20 Lines • Show All 89 Lines • ▼ Show 20 Lines
// CHECK: [[BPTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 0 // CHECK: [[BPTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 0
// CHECK: [[PTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 0 // CHECK: [[PTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 0
// CHECK: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 5, i8** [[BPTR]], i8** [[PTR]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZES1]], i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPTYPES1]], i32 0, i32 0), i8** null, i8** null) // CHECK: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 5, i8** [[BPTR]], i8** [[PTR]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZES1]], i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPTYPES1]], i32 0, i32 0), i8** null, i8** null)
// CHECK: foo // CHECK: foo
// %this.addr = alloca %struct.S*, align 8 // %this.addr = alloca %struct.S*, align 8
// CHECK: [[BPTRS:%.+]] = alloca [5 x i8*], // CHECK: [[BPTRS:%.+]] = alloca [5 x i8*],
// CHECK: [[PTRS:%.+]] = alloca [5 x i8*], // CHECK: [[PTRS:%.+]] = alloca [5 x i8*],
// CHECK: [[SIZES:%.+]] = alloca [5 x i64],
// %tmp = alloca i32*, align 8 // %tmp = alloca i32*, align 8
// %tmp6 = alloca i32**, align 8 // %tmp6 = alloca i32**, align 8
// %tmp7 = alloca i32*, align 8 // %tmp7 = alloca i32*, align 8
// %tmp8 = alloca i32**, align 8 // %tmp8 = alloca i32**, align 8
// %tmp9 = alloca [4 x i32]*, align 8 // %tmp9 = alloca [4 x i32]*, align 8
// store %struct.S* %this, %struct.S** %this.addr, align 8 // store %struct.S* %this, %struct.S** %this.addr, align 8
// %this1 = load %struct.S*, %struct.S** %this.addr, align 8 // %this1 = load %struct.S*, %struct.S** %this.addr, align 8
// CHECK: [[A_ADDR:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[THIS:%.+]], i32 0, i32 0 // CHECK: [[A_ADDR:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[THIS:%.+]], i32 0, i32 0
Show All 14 Lines
// CHECK: [[DIFF:%.+]] = sub i64 [[E]], [[B]] // CHECK: [[DIFF:%.+]] = sub i64 [[E]], [[B]]
// CHECK: [[SZ:%.+]] = sdiv exact i64 [[DIFF]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK: [[SZ:%.+]] = sdiv exact i64 [[DIFF]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK: [[BPTR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 0 // CHECK: [[BPTR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 0
// CHECK: [[BPTR0_S:%.+]] = bitcast i8** [[BPTR0]] to %struct.S** // CHECK: [[BPTR0_S:%.+]] = bitcast i8** [[BPTR0]] to %struct.S**
// CHECK: store %struct.S* [[THIS]], %struct.S** [[BPTR0_S]], // CHECK: store %struct.S* [[THIS]], %struct.S** [[BPTR0_S]],
// CHECK: [[PTR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 0 // CHECK: [[PTR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 0
// CHECK: [[PTR0_BEGIN:%.+]] = bitcast i8** [[PTR0]] to i32** // CHECK: [[PTR0_BEGIN:%.+]] = bitcast i8** [[PTR0]] to i32**
// CHECK: store i32* [[A_ADDR]], i32** [[PTR0_BEGIN]], // CHECK: store i32* [[A_ADDR]], i32** [[PTR0_BEGIN]],
// CHECK: [[SIZE0:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[SIZES]], i32 0, i32 0 // CHECK: store i64 [[SZ]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZES2]], i32 0, i32 0),
// CHECK: store i64 [[SZ]], i64* [[SIZE0]],
// CHECK: [[BPTR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 1 // CHECK: [[BPTR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 1
// CHECK: [[BPTR1_A_ADDR:%.+]] = bitcast i8** [[BPTR1]] to i32** // CHECK: [[BPTR1_A_ADDR:%.+]] = bitcast i8** [[BPTR1]] to i32**
// CHECK: store i32* [[A_ADDR2]], i32** [[BPTR1_A_ADDR]], // CHECK: store i32* [[A_ADDR2]], i32** [[BPTR1_A_ADDR]],
// CHECK: [[PTR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 1 // CHECK: [[PTR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 1
// CHECK: [[PTR1_A_ADDR:%.+]] = bitcast i8** [[PTR1]] to i32** // CHECK: [[PTR1_A_ADDR:%.+]] = bitcast i8** [[PTR1]] to i32**
// CHECK: store i32* [[A_ADDR2]], i32** [[PTR1_A_ADDR]], // CHECK: store i32* [[A_ADDR2]], i32** [[PTR1_A_ADDR]],
// CHECK: [[SIZE1:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[SIZES]], i32 0, i32 1
// CHECK: store i64 0, i64* [[SIZE1]],
// CHECK: [[BPTR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 2 // CHECK: [[BPTR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 2
// CHECK: [[BPTR2_PTR_ADDR:%.+]] = bitcast i8** [[BPTR2]] to i32*** // CHECK: [[BPTR2_PTR_ADDR:%.+]] = bitcast i8** [[BPTR2]] to i32***
// CHECK: store i32** [[PTR_ADDR]], i32*** [[BPTR2_PTR_ADDR]], // CHECK: store i32** [[PTR_ADDR]], i32*** [[BPTR2_PTR_ADDR]],
// CHECK: [[PTR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 2 // CHECK: [[PTR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 2
// CHECK: [[PTR2_PTR_ADDR:%.+]] = bitcast i8** [[PTR2]] to i32*** // CHECK: [[PTR2_PTR_ADDR:%.+]] = bitcast i8** [[PTR2]] to i32***
// CHECK: store i32** [[PTR_ADDR]], i32*** [[PTR2_PTR_ADDR]], // CHECK: store i32** [[PTR_ADDR]], i32*** [[PTR2_PTR_ADDR]],
// CHECK: [[SIZE2:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[SIZES]], i32 0, i32 2
// CHECK: store i64 0, i64* [[SIZE2]],
// CHECK: [[BPTR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 3 // CHECK: [[BPTR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 3
// CHECK: [[BPTR3_REF_PTR:%.+]] = bitcast i8** [[BPTR3]] to i32** // CHECK: [[BPTR3_REF_PTR:%.+]] = bitcast i8** [[BPTR3]] to i32**
// CHECK: store i32* [[REF_PTR]], i32** [[BPTR3_REF_PTR]], // CHECK: store i32* [[REF_PTR]], i32** [[BPTR3_REF_PTR]],
// CHECK: [[PTR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 3 // CHECK: [[PTR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 3
// CHECK: [[PTR3_REF_PTR:%.+]] = bitcast i8** [[PTR3]] to i32** // CHECK: [[PTR3_REF_PTR:%.+]] = bitcast i8** [[PTR3]] to i32**
// CHECK: store i32* [[REF_PTR]], i32** [[PTR3_REF_PTR]], // CHECK: store i32* [[REF_PTR]], i32** [[PTR3_REF_PTR]],
// CHECK: [[SIZE3:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[SIZES]], i32 0, i32 3
// CHECK: store i64 0, i64* [[SIZE3]],
// CHECK: [[BPTR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 4 // CHECK: [[BPTR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 4
// CHECK: [[BPTR4_ARR_ADDR:%.+]] = bitcast i8** [[BPTR4]] to [4 x i32]** // CHECK: [[BPTR4_ARR_ADDR:%.+]] = bitcast i8** [[BPTR4]] to [4 x i32]**
// CHECK: store [4 x i32]* [[ARR_ADDR2]], [4 x i32]** [[BPTR4_ARR_ADDR]], // CHECK: store [4 x i32]* [[ARR_ADDR2]], [4 x i32]** [[BPTR4_ARR_ADDR]],
// CHECK: [[PTR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 4 // CHECK: [[PTR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 4
// CHECK: [[PTR4_ARR_ADDR:%.+]] = bitcast i8** [[PTR4]] to [4 x i32]** // CHECK: [[PTR4_ARR_ADDR:%.+]] = bitcast i8** [[PTR4]] to [4 x i32]**
// CHECK: store [4 x i32]* [[ARR_ADDR2]], [4 x i32]** [[PTR4_ARR_ADDR]], // CHECK: store [4 x i32]* [[ARR_ADDR2]], [4 x i32]** [[PTR4_ARR_ADDR]],
// CHECK: [[SIZE4:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[SIZES]], i32 0, i32 4
// CHECK: store i64 0, i64* [[SIZE4]],
// CHECK: [[BPTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 0 // CHECK: [[BPTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 0
// CHECK: [[PTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 0 // CHECK: [[PTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 0
// CHECK: [[SIZE:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[SIZES]], i32 0, i32 0 // CHECK: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 5, i8** [[BPTR]], i8** [[PTR]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZES2]], i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPTYPES2]], i32 0, i32 0), i8** null, i8** null)
// CHECK: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 5, i8** [[BPTR]], i8** [[PTR]], i64* [[SIZE]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPTYPES2]], i32 0, i32 0), i8** null, i8** null)
// CHECK: [[A_ADDR:%.+]] = load i32*, i32** [[BPTR1_A_ADDR]], // CHECK: [[A_ADDR:%.+]] = load i32*, i32** [[BPTR1_A_ADDR]],
// CHECK: store i32* [[A_ADDR]], i32** [[A_REF:%.+]], // CHECK: store i32* [[A_ADDR]], i32** [[A_REF:%.+]],
// CHECK: [[PTR_ADDR:%.+]] = load i32**, i32*** [[BPTR2_PTR_ADDR]], // CHECK: [[PTR_ADDR:%.+]] = load i32**, i32*** [[BPTR2_PTR_ADDR]],
// CHECK: store i32** [[PTR_ADDR]], i32*** [[PTR_REF:%.+]], // CHECK: store i32** [[PTR_ADDR]], i32*** [[PTR_REF:%.+]],
// CHECK: [[REF_PTR:%.+]] = load i32*, i32** [[BPTR3_REF_PTR]], // CHECK: [[REF_PTR:%.+]] = load i32*, i32** [[BPTR3_REF_PTR]],
// CHECK: store i32* [[REF_PTR]], i32** [[REF_REF:%.+]], // CHECK: store i32* [[REF_PTR]], i32** [[REF_REF:%.+]],
// CHECK: [[PTR_ADDR:%.+]] = load i32**, i32*** [[BPTR2_PTR_ADDR]], // CHECK: [[PTR_ADDR:%.+]] = load i32**, i32*** [[BPTR2_PTR_ADDR]],
// CHECK: store i32** [[PTR_ADDR]], i32*** [[PTR_REF2:%.+]], // CHECK: store i32** [[PTR_ADDR]], i32*** [[PTR_REF2:%.+]],
Show All 14 Lines
// CHECK: store i32 [[INC]], i32* [[REF_PTR]], // CHECK: store i32 [[INC]], i32* [[REF_PTR]],
// CHECK: [[ARR_ADDR:%.+]] = load [4 x i32]*, [4 x i32]** [[ARR_REF]], // CHECK: [[ARR_ADDR:%.+]] = load [4 x i32]*, [4 x i32]** [[ARR_REF]],
// CHECK: [[ARR0_ADDR:%.+]] = getelementptr inbounds [4 x i32], [4 x i32]* [[ARR_ADDR]], i64 0, i64 0 // CHECK: [[ARR0_ADDR:%.+]] = getelementptr inbounds [4 x i32], [4 x i32]* [[ARR_ADDR]], i64 0, i64 0
// CHECK: [[VAL:%.+]] = load i32, i32* [[ARR0_ADDR]], // CHECK: [[VAL:%.+]] = load i32, i32* [[ARR0_ADDR]],
// CHECK: [[INC:%.+]] = add nsw i32 [[VAL]], 1 // CHECK: [[INC:%.+]] = add nsw i32 [[VAL]], 1
// CHECK: store i32 [[INC]], i32* [[ARR0_ADDR]], // CHECK: store i32 [[INC]], i32* [[ARR0_ADDR]],
// CHECK: [[BPTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 0 // CHECK: [[BPTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BPTRS]], i32 0, i32 0
// CHECK: [[PTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 0 // CHECK: [[PTR:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS]], i32 0, i32 0
// CHECK: [[SIZE:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[SIZES]], i32 0, i32 0 // CHECK: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 5, i8** [[BPTR]], i8** [[PTR]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZES2]], i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPTYPES2]], i32 0, i32 0), i8** null, i8** null)
// CHECK: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 5, i8** [[BPTR]], i8** [[PTR]], i64* [[SIZE]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPTYPES2]], i32 0, i32 0), i8** null, i8** null)
#endif #endif

clang/test/OpenMP/target_defaultmap_codegen_01.cpp

Show First 20 Lines • Show All 733 Lines • ▼ Show 20 Lines
// RUN: %clang_cc1 -DCK15 -verify -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY12 %s // RUN: %clang_cc1 -DCK15 -verify -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY12 %s
// RUN: %clang_cc1 -DCK15 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK15 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY12 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY12 %s
// SIMD-ONLY12-NOT: {{__kmpc|__tgt}} // SIMD-ONLY12-NOT: {{__kmpc|__tgt}}
#ifdef CK15 #ifdef CK15
// CK15-LABEL: @.__omp_offloading_{{.*}}implicit_maps_variable_length_array{{.*}}_l{{[0-9]+}}.region_id = weak{{.*}} constant i8 0 // CK15-LABEL: @.__omp_offloading_{{.*}}implicit_maps_variable_length_array{{.*}}_l{{[0-9]+}}.region_id = weak{{.*}} constant i8 0
// We don't have a constant map size for VLAs. // CK15-DAG: [[SIZES:@.+]] = {{.+}}global [3 x i64] [i64 {{4|8}}, i64 {{4|8}}, i64 0]
// Map types: // Map types:
// - OMP_MAP_LITERAL + OMP_MAP_TARGET_PARAM + OMP_MAP_IMPLICIT = 800 (vla size) // - OMP_MAP_LITERAL + OMP_MAP_TARGET_PARAM + OMP_MAP_IMPLICIT = 800 (vla size)
// - OMP_MAP_LITERAL + OMP_MAP_TARGET_PARAM + OMP_MAP_IMPLICIT = 800 (vla size) // - OMP_MAP_LITERAL + OMP_MAP_TARGET_PARAM + OMP_MAP_IMPLICIT = 800 (vla size)
// - OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 544 // - OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 544
// CK15-DAG: [[TYPES:@.+]] = {{.+}}constant [3 x i64] [i64 800, i64 800, i64 544] // CK15-DAG: [[TYPES:@.+]] = {{.+}}constant [3 x i64] [i64 800, i64 800, i64 544]
// CK15-LABEL: implicit_maps_variable_length_array{{.*}}( // CK15-LABEL: implicit_maps_variable_length_array{{.*}}(
void implicit_maps_variable_length_array (int a){ void implicit_maps_variable_length_array (int a){
double vla[2][a]; double vla[2][a];
// CK15-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{.+}}, i8* {{.+}}, i32 3, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], i64* [[SGEP:%[^,]+]], {{.+}}[[TYPES]]{{.+}}, i8** null, i8** null) // CK15-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{.+}}, i8* {{.+}}, i32 3, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}}, i8** null, i8** null)
// CK15-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0 // CK15-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
// CK15-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0 // CK15-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
// CK15-DAG: [[SGEP]] = getelementptr inbounds {{.+}}[[SS:%[^,]+]], i32 0, i32 0
// CK15-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0 // CK15-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
// CK15-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0 // CK15-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
// CK15-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[SS]], i32 0, i32 0
// CK15-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[sz:64|32]]* // CK15-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[sz:64|32]]*
// CK15-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[sz]]* // CK15-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[sz]]*
// CK15-DAG: store i[[sz]] 2, i[[sz]]* [[CBP0]] // CK15-DAG: store i[[sz]] 2, i[[sz]]* [[CBP0]]
// CK15-DAG: store i[[sz]] 2, i[[sz]]* [[CP0]] // CK15-DAG: store i[[sz]] 2, i[[sz]]* [[CP0]]
// CK15-DAG: store i64 {{8|4}}, i64* [[S0]],
// CK15-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1 // CK15-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1
// CK15-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1 // CK15-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1
// CK15-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[SS]], i32 0, i32 1
// CK15-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[sz]]* // CK15-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[sz]]*
// CK15-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[sz]]* // CK15-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[sz]]*
// CK15-DAG: store i[[sz]] [[VAL:%.+]], i[[sz]]* [[CBP1]] // CK15-DAG: store i[[sz]] [[VAL:%.+]], i[[sz]]* [[CBP1]]
// CK15-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP1]] // CK15-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP1]]
// CK15-DAG: store i64 {{8|4}}, i64* [[S1]],
// CK15-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2 // CK15-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2
// CK15-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2 // CK15-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2
// CK15-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[SS]], i32 0, i32 2
// CK15-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to double** // CK15-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to double**
// CK15-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double** // CK15-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double**
// CK15-DAG: store double* [[DECL:%.+]], double** [[CBP2]] // CK15-DAG: store double* [[DECL:%.+]], double** [[CBP2]]
// CK15-DAG: store double* [[DECL]], double** [[CP2]] // CK15-DAG: store double* [[DECL]], double** [[CP2]]
// CK15-DAG: store i64 [[VALS2:%.+]], i64* [[S2]], // CK15-DAG: store i64 [[VALS2:%.+]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZES]], i32 0, i32 2),
// CK15-DAG: [[VALS2]] = {{mul nuw i64 %.+, 8|sext i32 %.+ to i64}} // CK15-DAG: [[VALS2]] = {{mul nuw i64 %.+, 8|sext i32 %.+ to i64}}
// CK15: call void [[KERNEL:@.+]](i[[sz]] {{.+}}, i[[sz]] {{.+}}, double* [[DECL]]) // CK15: call void [[KERNEL:@.+]](i[[sz]] {{.+}}, i[[sz]] {{.+}}, double* [[DECL]])
#pragma omp target defaultmap(alloc: aggregate) #pragma omp target defaultmap(alloc: aggregate)
{ {
vla[1][3] += 1.0; vla[1][3] += 1.0;
} }
} }
▲ Show 20 Lines • Show All 766 Lines • Show Last 20 Lines

clang/test/OpenMP/target_enter_data_codegen.cpp

Show All 38 Lines
// CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i64] [i64 800] // CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i64] [i64 800]
// CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] zeroinitializer // CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] zeroinitializer
// CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i64] [i64 4] // CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i64] [i64 4]
// CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 1] // CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 1]
// CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 5] // CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 5]
// CK1: [[SIZE04:@.+]] = {{.+}}constant [2 x i64] [i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64 24] // CK1: [[SIZE04:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673] // CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673]
// CK1: [[MTYPE05:@.+]] = {{.+}}constant [1 x i64] [i64 1025] // CK1: [[MTYPE05:@.+]] = {{.+}}constant [1 x i64] [i64 1025]
// CK1: [[MTYPE06:@.+]] = {{.+}}constant [1 x i64] [i64 1029] // CK1: [[MTYPE06:@.+]] = {{.+}}constant [1 x i64] [i64 1029]
// CK1-LABEL: _Z3fooi // CK1-LABEL: _Z3fooi
void foo(int arg) { void foo(int arg) {
▲ Show 20 Lines • Show All 95 Lines • ▼ Show 20 Linesvoid foo(int arg) {
// CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK1-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK1-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK1-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK1-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK1-DAG: store [[ST]]* @gb, [[ST]]** [[CBP0]] // CK1-DAG: store [[ST]]* @gb, [[ST]]** [[CBP0]]
// CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CP0]] // CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CP0]]
// CK1-DAG: store i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE04]], i32 0, i32 0),
// CK1-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK1-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK1-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK1-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK1-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK1-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK1-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK1-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CBP1]] // CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CBP1]]
// CK1-DAG: store double* [[SEC1:%.+]], double** [[CP1]] // CK1-DAG: store double* [[SEC1:%.+]], double** [[CP1]]
▲ Show 20 Lines • Show All 183 Lines • ▼ Show 20 Linesstruct ST {
T foo(T arg) { T foo(T arg) {
// Region 00 // Region 00
#pragma omp target enter data map(always, to: b[1:3]) if(a>123) device(arg) #pragma omp target enter data map(always, to: b[1:3]) if(a>123) device(arg)
{arg++;} {arg++;}
return arg; return arg;
} }
}; };
// CK2: [[SIZES:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK2: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710677] // CK2: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710677]
// CK2-LABEL: _Z3bari // CK2-LABEL: _Z3bari
int bar(int arg){ int bar(int arg){
ST<int> A; ST<int> A;
return A.foo(arg); return A.foo(arg);
} }
// Region 00 // Region 00
// CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] // CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
// CK2: [[IFTHEN]] // CK2: [[IFTHEN]]
// CK2-DAG: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[sz:64|32]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK2-DAG: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZES]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK2-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64 // CK2-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64
// CK2-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}}, // CK2-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}},
// CK2-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK2-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK2-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK2-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK2-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK2-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK2-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK2-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK2-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK2-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK2-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK2-DAG: store [[ST]]* [[VAR0:%.+]], [[ST]]** [[CBP0]] // CK2-DAG: store [[ST]]* [[VAR0:%.+]], [[ST]]** [[CBP0]]
// CK2-DAG: store double** [[SEC0:%.+]], double*** [[CP0]] // CK2-DAG: store double** [[SEC0:%.+]], double*** [[CP0]]
// CK2-DAG: store i64 {{%.+}}, i64* [[S0]] // CK2-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZES]], i32 0, i32 0),
// CK2-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1 // CK2-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1
// CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK2-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK2-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK2-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK2-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK2-DAG: store double** [[SEC0]], double*** [[CBP1]] // CK2-DAG: store double** [[SEC0]], double*** [[CBP1]]
// CK2-DAG: store double* [[SEC1:%.+]], double** [[CP1]] // CK2-DAG: store double* [[SEC1:%.+]], double** [[CP1]]
▲ Show 20 Lines • Show All 108 Lines • ▼ Show 20 Linesstruct STT {
T foo(T arg) { T foo(T arg) {
// Region 00 // Region 00
#pragma omp target enter data map(always close to: b[1:3]) if(a>123) device(arg) #pragma omp target enter data map(always close to: b[1:3]) if(a>123) device(arg)
{arg++;} {arg++;}
return arg; return arg;
} }
}; };
// CK5: [[SIZES:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK5: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976711701] // CK5: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976711701]
// CK5-LABEL: _Z3bari // CK5-LABEL: _Z3bari
int bar(int arg){ int bar(int arg){
STT<int> A; STT<int> A;
return A.foo(arg); return A.foo(arg);
} }
// Region 00 // Region 00
// CK5: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] // CK5: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
// CK5: [[IFTHEN]] // CK5: [[IFTHEN]]
// CK5-DAG: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[sz:64|32]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK5-DAG: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZES]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK5-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64 // CK5-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64
// CK5-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}}, // CK5-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}},
// CK5-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK5-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK5-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK5-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK5-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK5-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK5-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK5-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK5-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK5-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK5-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[STT]]** // CK5-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[STT]]**
// CK5-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK5-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK5-DAG: store [[STT]]* [[VAR0:%.+]], [[STT]]** [[CBP0]] // CK5-DAG: store [[STT]]* [[VAR0:%.+]], [[STT]]** [[CBP0]]
// CK5-DAG: store double** [[SEC0:%.+]], double*** [[CP0]] // CK5-DAG: store double** [[SEC0:%.+]], double*** [[CP0]]
// CK5-DAG: store i64 {{%.+}}, i64* [[S0]] // CK5-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZES]], i32 0, i32 0),
// CK5-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[STT]]* [[VAR0]], i32 0, i32 1 // CK5-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[STT]]* [[VAR0]], i32 0, i32 1
// CK5-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK5-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK5-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK5-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK5-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK5-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK5-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK5-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK5-DAG: store double** [[SEC0]], double*** [[CBP1]] // CK5-DAG: store double** [[SEC0]], double*** [[CBP1]]
// CK5-DAG: store double* [[SEC1:%.+]], double** [[CP1]] // CK5-DAG: store double* [[SEC1:%.+]], double** [[CP1]]
Show All 13 Lines

clang/test/OpenMP/target_enter_data_depend_codegen.cpp

Show All 31 Lines
// CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i64] [i64 800] // CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i64] [i64 800]
// CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] zeroinitializer // CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] zeroinitializer
// CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i64] [i64 4] // CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i64] [i64 4]
// CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 1] // CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 1]
// CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] zeroinitializer // CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] zeroinitializer
// CK1: [[SIZE04:@.+]] = {{.+}}constant [2 x i64] [i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64 24] // CK1: [[SIZE04:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673] // CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673]
// CK1-LABEL: _Z3fooi // CK1-LABEL: _Z3fooi
void foo(int arg) { void foo(int arg) {
int la; int la;
float lb[arg]; float lb[arg];
// CK1: alloca [1 x %struct.kmp_depend_info], // CK1: alloca [1 x %struct.kmp_depend_info],
▲ Show 20 Lines • Show All 236 Lines • ▼ Show 20 Linesvoid foo(int arg) {
// Region 04 // Region 04
// CK1: [[BP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%.+]], i32 0, i32 0 // CK1: [[BP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%.+]], i32 0, i32 0
// CK1: [[BP0_BC:%.+]] = bitcast i8** [[BP0]] to %struct.ST** // CK1: [[BP0_BC:%.+]] = bitcast i8** [[BP0]] to %struct.ST**
// CK1: store %struct.ST* @gb, %struct.ST** [[BP0_BC]], // CK1: store %struct.ST* @gb, %struct.ST** [[BP0_BC]],
// CK1: [[P0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%.+]], i32 0, i32 0 // CK1: [[P0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%.+]], i32 0, i32 0
// CK1: [[P0_BC:%.+]] = bitcast i8** [[P0]] to double*** // CK1: [[P0_BC:%.+]] = bitcast i8** [[P0]] to double***
// CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[P0_BC]], // CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[P0_BC]],
// CK1: store i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE04]], i32 0, i32 0),
// CK1: [[BP1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 1 // CK1: [[BP1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 1
// CK1: [[BP1_BC:%.+]] = bitcast i8** [[BP1]] to double*** // CK1: [[BP1_BC:%.+]] = bitcast i8** [[BP1]] to double***
// CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[BP1_BC]], // CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[BP1_BC]],
// CK1: [[P1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 1 // CK1: [[P1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 1
// CK1: [[P1_BC:%.+]] = bitcast i8** [[P1]] to double** // CK1: [[P1_BC:%.+]] = bitcast i8** [[P1]] to double**
// CK1: store double* %{{.+}}, double** [[P1_BC]], // CK1: store double* %{{.+}}, double** [[P1_BC]],
// CK1: [[GEPBP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 0 // CK1: [[GEPBP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 0
// CK1: [[GEPP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 0 // CK1: [[GEPP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 0
▲ Show 20 Lines • Show All 131 Lines • Show Last 20 Lines

clang/test/OpenMP/target_exit_data_codegen.cpp

Show All 37 Lines
// CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i64] [i64 800] // CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i64] [i64 800]
// CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] [i64 2] // CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] [i64 2]
// CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i64] [i64 4] // CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i64] [i64 4]
// CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] zeroinitializer // CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] zeroinitializer
// CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 6] // CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 6]
// CK1: [[SIZE04:@.+]] = {{.+}}constant [2 x i64] [i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64 24] // CK1: [[SIZE04:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710672] // CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710672]
// CK1: [[MTYPE05:@.+]] = {{.+}}constant [1 x i64] [i64 1026] // CK1: [[MTYPE05:@.+]] = {{.+}}constant [1 x i64] [i64 1026]
// CK1: [[MTYPE06:@.+]] = {{.+}}constant [1 x i64] [i64 1030] // CK1: [[MTYPE06:@.+]] = {{.+}}constant [1 x i64] [i64 1030]
// CK1-LABEL: _Z3fooi // CK1-LABEL: _Z3fooi
void foo(int arg) { void foo(int arg) {
▲ Show 20 Lines • Show All 95 Lines • ▼ Show 20 Linesvoid foo(int arg) {
// CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK1-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK1-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK1-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK1-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK1-DAG: store [[ST]]* @gb, [[ST]]** [[CBP0]] // CK1-DAG: store [[ST]]* @gb, [[ST]]** [[CBP0]]
// CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CP0]] // CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CP0]]
// CK1-DAG: store i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE04]], i32 0, i32 0),
// CK1-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK1-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK1-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK1-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK1-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK1-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK1-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK1-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CBP1]] // CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CBP1]]
// CK1-DAG: store double* [[SEC1:%[^,]+]], double** [[CP1]] // CK1-DAG: store double* [[SEC1:%[^,]+]], double** [[CP1]]
▲ Show 20 Lines • Show All 95 Lines • ▼ Show 20 Linesstruct ST {
T foo(T arg) { T foo(T arg) {
// Region 00 // Region 00
#pragma omp target exit data map(always, release: b[1:3]) if(a>123) device(arg) #pragma omp target exit data map(always, release: b[1:3]) if(a>123) device(arg)
{arg++;} {arg++;}
return arg; return arg;
} }
}; };
// CK2: [[SIZES:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK2: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710676] // CK2: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710676]
// CK2-LABEL: _Z3bari // CK2-LABEL: _Z3bari
int bar(int arg){ int bar(int arg){
ST<int> A; ST<int> A;
return A.foo(arg); return A.foo(arg);
} }
// Region 00 // Region 00
// CK2-NOT: __tgt_target_data_begin // CK2-NOT: __tgt_target_data_begin
// CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] // CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
// CK2: [[IFTHEN]] // CK2: [[IFTHEN]]
// CK2-DAG: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[sz:.+]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK2-DAG: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZES]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK2-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64 // CK2-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64
// CK2-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}}, // CK2-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}},
// CK2-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK2-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK2-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK2-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK2-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK2-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK2-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK2-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK2-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK2-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK2-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK2-DAG: store [[ST]]* [[VAR0:%[^,]+]], [[ST]]** [[CBP0]] // CK2-DAG: store [[ST]]* [[VAR0:%[^,]+]], [[ST]]** [[CBP0]]
// CK2-DAG: store double** [[SEC0:%[^,]+]], double*** [[CP0]] // CK2-DAG: store double** [[SEC0:%[^,]+]], double*** [[CP0]]
// CK2-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]] // CK2-DAG: store i64 [[CSVAL0:%[^,]+]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZES]], i32 0, i32 0),
// CK2-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1 // CK2-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1
// CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK2-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK2-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK2-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK2-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK2-DAG: store double** [[SEC0]], double*** [[CBP1]] // CK2-DAG: store double** [[SEC0]], double*** [[CBP1]]
// CK2-DAG: store double* [[SEC1:%[^,]+]], double** [[CP1]] // CK2-DAG: store double* [[SEC1:%[^,]+]], double** [[CP1]]
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Linesstruct STT {
T foo(T arg) { T foo(T arg) {
// Region 00 // Region 00
#pragma omp target exit data map(always close, release: b[1:3]) if(a>123) device(arg) #pragma omp target exit data map(always close, release: b[1:3]) if(a>123) device(arg)
{arg++;} {arg++;}
return arg; return arg;
} }
}; };
// CK4: [[SIZES:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK4: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976711700] // CK4: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976711700]
// CK4-LABEL: _Z3bari // CK4-LABEL: _Z3bari
int bar(int arg){ int bar(int arg){
STT<int> A; STT<int> A;
return A.foo(arg); return A.foo(arg);
} }
// Region 00 // Region 00
// CK4-NOT: __tgt_target_data_begin // CK4-NOT: __tgt_target_data_begin
// CK4: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] // CK4: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
// CK4: [[IFTHEN]] // CK4: [[IFTHEN]]
// CK4-DAG: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[sz:.+]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK4-DAG: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZES]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK4-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64 // CK4-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64
// CK4-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}}, // CK4-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}},
// CK4-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK4-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK4-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK4-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK4-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK4-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK4-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK4-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK4-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK4-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK4-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[STT]]** // CK4-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[STT]]**
// CK4-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK4-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK4-DAG: store [[STT]]* [[VAR0:%[^,]+]], [[STT]]** [[CBP0]] // CK4-DAG: store [[STT]]* [[VAR0:%[^,]+]], [[STT]]** [[CBP0]]
// CK4-DAG: store double** [[SEC0:%[^,]+]], double*** [[CP0]] // CK4-DAG: store double** [[SEC0:%[^,]+]], double*** [[CP0]]
// CK4-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]] // CK4-DAG: store i64 [[CSVAL0:%[^,]+]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZES]], i32 0, i32 0),
// CK4-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[STT]]* [[VAR0]], i32 0, i32 1 // CK4-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[STT]]* [[VAR0]], i32 0, i32 1
// CK4-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK4-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK4-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK4-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK4-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK4-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK4-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK4-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK4-DAG: store double** [[SEC0]], double*** [[CBP1]] // CK4-DAG: store double** [[SEC0]], double*** [[CBP1]]
// CK4-DAG: store double* [[SEC1:%[^,]+]], double** [[CP1]] // CK4-DAG: store double* [[SEC1:%[^,]+]], double** [[CP1]]
Show All 12 Lines

clang/test/OpenMP/target_exit_data_depend_codegen.cpp

Show All 31 Lines
// CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i64] [i64 800] // CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i64] [i64 800]
// CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] [i64 2] // CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] [i64 2]
// CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i64] [i64 4] // CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i64] [i64 4]
// CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 8] // CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 8]
// CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 2] // CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 2]
// CK1: [[SIZE04:@.+]] = {{.+}}constant [2 x i64] [i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64 24] // CK1: [[SIZE04:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710674] // CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710674]
// CK1-LABEL: _Z3fooi // CK1-LABEL: _Z3fooi
void foo(int arg) { void foo(int arg) {
int la; int la;
float lb[arg]; float lb[arg];
// CK1: alloca [1 x %struct.kmp_depend_info], // CK1: alloca [1 x %struct.kmp_depend_info],
▲ Show 20 Lines • Show All 236 Lines • ▼ Show 20 Linesvoid foo(int arg) {
// Region 04 // Region 04
// CK1: [[BP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%.+]], i32 0, i32 0 // CK1: [[BP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%.+]], i32 0, i32 0
// CK1: [[BP0_BC:%.+]] = bitcast i8** [[BP0]] to %struct.ST** // CK1: [[BP0_BC:%.+]] = bitcast i8** [[BP0]] to %struct.ST**
// CK1: store %struct.ST* @gb, %struct.ST** [[BP0_BC]], // CK1: store %struct.ST* @gb, %struct.ST** [[BP0_BC]],
// CK1: [[P0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%.+]], i32 0, i32 0 // CK1: [[P0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%.+]], i32 0, i32 0
// CK1: [[P0_BC:%.+]] = bitcast i8** [[P0]] to double*** // CK1: [[P0_BC:%.+]] = bitcast i8** [[P0]] to double***
// CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[P0_BC]], // CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[P0_BC]],
// CK1: store i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE04]], i32 0, i32 0),
// CK1: [[BP1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 1 // CK1: [[BP1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 1
// CK1: [[BP1_BC:%.+]] = bitcast i8** [[BP1]] to double*** // CK1: [[BP1_BC:%.+]] = bitcast i8** [[BP1]] to double***
// CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[BP1_BC]], // CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[BP1_BC]],
// CK1: [[P1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 1 // CK1: [[P1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 1
// CK1: [[P1_BC:%.+]] = bitcast i8** [[P1]] to double** // CK1: [[P1_BC:%.+]] = bitcast i8** [[P1]] to double**
// CK1: store double* %{{.+}}, double** [[P1_BC]], // CK1: store double* %{{.+}}, double** [[P1_BC]],
// CK1: [[GEPBP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 0 // CK1: [[GEPBP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 0
// CK1: [[GEPP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 0 // CK1: [[GEPP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 0
▲ Show 20 Lines • Show All 131 Lines • Show Last 20 Lines

clang/test/OpenMP/target_firstprivate_codegen.cpp

Show First 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK-DAG: [[S1:%.+]] = type { double } // CHECK-DAG: [[S1:%.+]] = type { double }
// TCHECK-DAG: [[TT:%.+]] = type { i64, i8 } // TCHECK-DAG: [[TT:%.+]] = type { i64, i8 }
// TCHECK-DAG: [[TTII:%.+]] = type { i32, i32 } // TCHECK-DAG: [[TTII:%.+]] = type { i32, i32 }
// TCHECK-DAG: [[S1:%.+]] = type { double } // TCHECK-DAG: [[S1:%.+]] = type { double }
// CHECK-DAG: [[SIZET:@.+]] = private unnamed_addr constant [3 x i{{32|64}}] [i[[SZ:32|64]] 4, i{{64|32}} {{8|4}}, i[[SZ:32|64]] 4] // CHECK-DAG: [[SIZET:@.+]] = private unnamed_addr constant [3 x i{{32|64}}] [i[[SZ:32|64]] 4, i{{64|32}} {{8|4}}, i[[SZ:32|64]] 4]
// CHECK-DAG: [[MAPT:@.+]] = private unnamed_addr constant [3 x i64] [i64 288, i64 49, i64 288] // CHECK-DAG: [[MAPT:@.+]] = private unnamed_addr constant [3 x i64] [i64 288, i64 49, i64 288]
// CHECK-DAG: [[SIZET2:@.+]] = private unnamed_addr global [9 x i64] [i64 2, i64 40, i64 {{4|8}}, i64 0, i64 400, i64 {{4|8}}, i64 {{4|8}}, i64 0, i64 {{12|16}}]
// CHECK-DAG: [[MAPT2:@.+]] = private unnamed_addr constant [9 x i64] [i64 288, i64 161, i64 800, i64 161, i64 161, i64 800, i64 800, i64 161, i64 161] // CHECK-DAG: [[MAPT2:@.+]] = private unnamed_addr constant [9 x i64] [i64 288, i64 161, i64 800, i64 161, i64 161, i64 800, i64 800, i64 161, i64 161]
// CHECK-DAG: [[SIZET3:@.+]] = private unnamed_addr constant [2 x i{{32|64}}] [i{{32|64}} 0, i{{32|64}} 8] // CHECK-DAG: [[SIZET3:@.+]] = private unnamed_addr constant [2 x i{{32|64}}] [i{{32|64}} 0, i{{32|64}} 8]
// CHECK-DAG: [[MAPT3:@.+]] = private unnamed_addr constant [2 x i64] [i64 32, i64 161] // CHECK-DAG: [[MAPT3:@.+]] = private unnamed_addr constant [2 x i64] [i64 32, i64 161]
// CHECK-DAG: [[SIZET4:@.+]] = private unnamed_addr global [5 x i64] [i64 8, i64 4, i64 {{4|8}}, i64 {{4|8}}, i64 0]
// CHECK-DAG: [[MAPT4:@.+]] = private unnamed_addr constant [5 x i64] [i64 547, i64 288, i64 800, i64 800, i64 161] // CHECK-DAG: [[MAPT4:@.+]] = private unnamed_addr constant [5 x i64] [i64 547, i64 288, i64 800, i64 800, i64 161]
// CHECK-DAG: [[SIZET5:@.+]] = private unnamed_addr constant [3 x i{{32|64}}] [i[[SZ]] 4, i[[SZ]] 1, i[[SZ]] 40] // CHECK-DAG: [[SIZET5:@.+]] = private unnamed_addr constant [3 x i{{32|64}}] [i[[SZ]] 4, i[[SZ]] 1, i[[SZ]] 40]
// CHECK-DAG: [[MAPT5:@.+]] = private unnamed_addr constant [3 x i64] [i64 288, i64 288, i64 161] // CHECK-DAG: [[MAPT5:@.+]] = private unnamed_addr constant [3 x i64] [i64 288, i64 288, i64 161]
// CHECK-DAG: [[SIZET6:@.+]] = private unnamed_addr constant [2 x i{{32|64}}] [i[[SZ]] 4, i[[SZ]] 40] // CHECK-DAG: [[SIZET6:@.+]] = private unnamed_addr constant [2 x i{{32|64}}] [i[[SZ]] 4, i[[SZ]] 40]
// CHECK-DAG: [[MAPT6:@.+]] = private unnamed_addr constant [2 x i64] [i64 288, i64 161] // CHECK-DAG: [[MAPT6:@.+]] = private unnamed_addr constant [2 x i64] [i64 288, i64 161]
// CHECK: define {{.*}}[[FOO:@.+]]( // CHECK: define {{.*}}[[FOO:@.+]](
int foo(int n, double *ptr) { int foo(int n, double *ptr) {
Show All 23 Lines#pragma omp target firstprivate(a, p, ga)
// CHECK: [[FP_E:%.+]] = alloca [[TTII]], // CHECK: [[FP_E:%.+]] = alloca [[TTII]],
// CHECK: [[P:%.+]] = alloca i32*, align 64 // CHECK: [[P:%.+]] = alloca i32*, align 64
// CHECK: [[ACAST:%.+]] = alloca i{{[0-9]+}}, // CHECK: [[ACAST:%.+]] = alloca i{{[0-9]+}},
// CHECK: [[BASE_PTR_ARR:%.+]] = alloca [3 x i8*], // CHECK: [[BASE_PTR_ARR:%.+]] = alloca [3 x i8*],
// CHECK: [[PTR_ARR:%.+]] = alloca [3 x i8*], // CHECK: [[PTR_ARR:%.+]] = alloca [3 x i8*],
// CHECK: [[A2CAST:%.+]] = alloca i{{[0-9]+}}, // CHECK: [[A2CAST:%.+]] = alloca i{{[0-9]+}},
// CHECK: [[BASE_PTR_ARR2:%.+]] = alloca [9 x i8*], // CHECK: [[BASE_PTR_ARR2:%.+]] = alloca [9 x i8*],
// CHECK: [[PTR_ARR2:%.+]] = alloca [9 x i8*], // CHECK: [[PTR_ARR2:%.+]] = alloca [9 x i8*],
// CHECK: [[SIZET2:%.+]] = alloca [9 x i{{[0-9]+}}],
// CHECK: [[BASE_PTR_ARR3:%.+]] = alloca [2 x i8*], // CHECK: [[BASE_PTR_ARR3:%.+]] = alloca [2 x i8*],
// CHECK: [[PTR_ARR3:%.+]] = alloca [2 x i8*], // CHECK: [[PTR_ARR3:%.+]] = alloca [2 x i8*],
// CHECK: [[N_ADDR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[N_ADDR]], // CHECK: [[N_ADDR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[N_ADDR]],
// CHECK-64: [[N_EXT:%.+]] = zext i{{[0-9]+}} [[N_ADDR_VAL]] to i{{[0-9]+}} // CHECK-64: [[N_EXT:%.+]] = zext i{{[0-9]+}} [[N_ADDR_VAL]] to i{{[0-9]+}}
// CHECK: [[SSAVE_RET:%.+]] = call i8* @llvm.stacksave() // CHECK: [[SSAVE_RET:%.+]] = call i8* @llvm.stacksave()
// CHECK: store i8* [[SSAVE_RET]], i8** [[SSTACK]], // CHECK: store i8* [[SSAVE_RET]], i8** [[SSTACK]],
// CHECK-64: [[BN_VLA:%.+]] = alloca float, i{{[0-9]+}} [[N_EXT]], // CHECK-64: [[BN_VLA:%.+]] = alloca float, i{{[0-9]+}} [[N_EXT]],
// CHECK-32: [[BN_VLA:%.+]] = alloca float, i{{[0-9]+}} [[N_ADDR_VAL]], // CHECK-32: [[BN_VLA:%.+]] = alloca float, i{{[0-9]+}} [[N_ADDR_VAL]],
▲ Show 20 Lines • Show All 68 Lines • ▼ Show 20 Lines#pragma omp target firstprivate(aa, b, bn, c, cn, d)
// firstprivate(aa) --> base_ptr = aa, ptr = aa, size = 2 (short) // firstprivate(aa) --> base_ptr = aa, ptr = aa, size = 2 (short)
// CHECK: [[BASE_PTR_GEP2_0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 // CHECK: [[BASE_PTR_GEP2_0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: [[ACAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_0]] to i{{[0-9]+}}* // CHECK: [[ACAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_0]] to i{{[0-9]+}}*
// CHECK: store i{{[0-9]+}} [[A2CAST_VAL]], i{{[0-9]+}}* [[ACAST_TOPTR]], // CHECK: store i{{[0-9]+}} [[A2CAST_VAL]], i{{[0-9]+}}* [[ACAST_TOPTR]],
// CHECK: [[PTR_GEP2_0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 // CHECK: [[PTR_GEP2_0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: [[ACAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_0]] to i{{[0-9]+}}* // CHECK: [[ACAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_0]] to i{{[0-9]+}}*
// CHECK: store i{{[0-9]+}} [[A2CAST_VAL]], i{{[0-9]+}}* [[ACAST_TOPTR]], // CHECK: store i{{[0-9]+}} [[A2CAST_VAL]], i{{[0-9]+}}* [[ACAST_TOPTR]],
// CHECK: [[SIZE_GEPA2:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIZE_GEPA2]],
// firstprivate(b): base_ptr = &b[0], ptr = &b[0], size = 40 (sizeof(float)*10) // firstprivate(b): base_ptr = &b[0], ptr = &b[0], size = 40 (sizeof(float)*10)
// CHECK: [[BASE_PTR_GEP2_1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 // CHECK: [[BASE_PTR_GEP2_1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_1]] to [10 x float]** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_1]] to [10 x float]**
// CHECK: store [10 x float]* [[B]], [10 x float]** [[BCAST_TOPTR]], // CHECK: store [10 x float]* [[B]], [10 x float]** [[BCAST_TOPTR]],
// CHECK: [[PTR_GEP2_1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 // CHECK: [[PTR_GEP2_1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_1]] to [10 x float]** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_1]] to [10 x float]**
// CHECK: store [10 x float]* [[B]], [10 x float]** [[BCAST_TOPTR]], // CHECK: store [10 x float]* [[B]], [10 x float]** [[BCAST_TOPTR]],
// CHECK: [[SIZE_GEPB:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
// CHECK: store i{{[0-9]+}} 40, i{{[0-9]+}}* [[SIZE_GEPB]],
// firstprivate(bn), 2 entries, n and bn: (1) base_ptr = n, ptr = n, size = 8 ; (2) base_ptr = &c[0], ptr = &c[0], size = n*sizeof(float) // firstprivate(bn), 2 entries, n and bn: (1) base_ptr = n, ptr = n, size = 8 ; (2) base_ptr = &c[0], ptr = &c[0], size = n*sizeof(float)
// CHECK: [[BASE_PTR_GEP2_2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 // CHECK: [[BASE_PTR_GEP2_2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_2]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_2]] to i{{[0-9]+}}*
// CHECK-64: store i{{[0-9]+}} [[N_EXT]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK-64: store i{{[0-9]+}} [[N_EXT]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[PTR_GEP2_2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 // CHECK: [[PTR_GEP2_2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_2]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_2]] to i{{[0-9]+}}*
// CHECK-64: store i{{[0-9]+}} [[N_EXT]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK-64: store i{{[0-9]+}} [[N_EXT]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[SIZE_GEPBN_1:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
// CHECK: store i{{[0-9]+}} {{[0-9]}}, i{{[0-9]+}}* [[SIZE_GEPBN_1]],
// CHECK: [[BASE_PTR_GEP2_3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 3 // CHECK: [[BASE_PTR_GEP2_3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_3]] to float** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_3]] to float**
// CHECK: store float* [[BN_VLA]], float** [[BCAST_TOPTR]], // CHECK: store float* [[BN_VLA]], float** [[BCAST_TOPTR]],
// CHECK: [[SIZE_GEPBN_3:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 3 // CHECK: store i{{[0-9]+}} [[BN_SIZE]], i{{[0-9]+}}* getelementptr inbounds ([9 x i64], [9 x i64]* [[SIZET2]], i32 0, i32 3),
// CHECK: store i{{[0-9]+}} [[BN_SIZE]], i{{[0-9]+}}* [[SIZE_GEPBN_3]]
// firstprivate(c): base_ptr = &c[0], ptr = &c[0], size = 400 (5*10*sizeof(double)) // firstprivate(c): base_ptr = &c[0], ptr = &c[0], size = 400 (5*10*sizeof(double))
// CHECK: [[BASE_PTR_GEP2_4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 4 // CHECK: [[BASE_PTR_GEP2_4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_4]] to [5 x [10 x double]]** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_4]] to [5 x [10 x double]]**
// CHECK: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[BCAST_TOPTR]], // CHECK: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[BCAST_TOPTR]],
// CHECK: [[PTR_GEP2_4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 4 // CHECK: [[PTR_GEP2_4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_4]] to [5 x [10 x double]]** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_4]] to [5 x [10 x double]]**
// CHECK: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[BCAST_TOPTR]], // CHECK: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[BCAST_TOPTR]],
// CHECK: [[SIZE_GEPC_4:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
// CHECK: store i{{[0-9]+}} 400, i{{[0-9]+}}* [[SIZE_GEPC_4]],
// firstprivate(cn), 3 entries, 5, n, cn: (1) base_ptr = 5, ptr = 5, size = 8; (2) (1) base_ptr = n, ptr = n, size = 8; (3) base_ptr = &cn[0], ptr = &cn[0], size = 5*n*sizeof(double) // firstprivate(cn), 3 entries, 5, n, cn: (1) base_ptr = 5, ptr = 5, size = 8; (2) (1) base_ptr = n, ptr = n, size = 8; (3) base_ptr = &cn[0], ptr = &cn[0], size = 5*n*sizeof(double)
// CHECK: [[BASE_PTR_GEP2_5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 5 // CHECK: [[BASE_PTR_GEP2_5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 5
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_5]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_5]] to i{{[0-9]+}}*
// CHECK: store i{{[0-9]+}} 5, i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK: store i{{[0-9]+}} 5, i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[PTR_GEP2_5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 5 // CHECK: [[PTR_GEP2_5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 5
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_5]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_5]] to i{{[0-9]+}}*
// CHECK: store i{{[0-9]+}} 5, i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK: store i{{[0-9]+}} 5, i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[SIZE_GEPCN_5:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 5
// CHECK: store i{{[0-9]+}} {{[0-9]}}, i{{[0-9]+}}* [[SIZE_GEPCN_5]],
// CHECK: [[BASE_PTR_GEP2_6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 6 // CHECK: [[BASE_PTR_GEP2_6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 6
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_6]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_6]] to i{{[0-9]+}}*
// CHECK-64: store i{{[0-9]+}} [[N_EXT2]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK-64: store i{{[0-9]+}} [[N_EXT2]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL2]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL2]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[PTR_GEP2_6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 6 // CHECK: [[PTR_GEP2_6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 6
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_6]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_6]] to i{{[0-9]+}}*
// CHECK-64: store i{{[0-9]+}} [[N_EXT2]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK-64: store i{{[0-9]+}} [[N_EXT2]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL2]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL2]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[SIZE_GEPCN_6:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 6
// CHECK: store i{{[0-9]+}} {{[0-9]}}, i{{[0-9]+}}* [[SIZE_GEPCN_6]],
// CHECK: [[BASE_PTR_GEP2_7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 7 // CHECK: [[BASE_PTR_GEP2_7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 7
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_7]] to double** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_7]] to double**
// CHECK: store double* [[CN_VLA]], double** [[BCAST_TOPTR]], // CHECK: store double* [[CN_VLA]], double** [[BCAST_TOPTR]],
// CHECK: [[PTR_GEP2_7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 7 // CHECK: [[PTR_GEP2_7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 7
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_7]] to double** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_7]] to double**
// CHECK: store double* [[CN_VLA]], double** [[BCAST_TOPTR]], // CHECK: store double* [[CN_VLA]], double** [[BCAST_TOPTR]],
// CHECK: [[SIZE_GEPCN_7:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 7 // CHECK: store i{{[0-9]+}} [[CN_SIZE_2]], i{{[0-9]+}}* getelementptr inbounds ([9 x i64], [9 x i64]* [[SIZET2]], i32 0, i32 7),
// CHECK: store i{{[0-9]+}} [[CN_SIZE_2]], i{{[0-9]+}}* [[SIZE_GEPCN_7]],
// firstprivate(d): base_ptr = &d, ptr = &d, size = 16 // firstprivate(d): base_ptr = &d, ptr = &d, size = 16
// CHECK: [[BASE_PTR_GEP2_8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 8 // CHECK: [[BASE_PTR_GEP2_8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 8
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_8]] to [[TT]]** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_8]] to [[TT]]**
// CHECK: store [[TT]]* [[D]], [[TT]]** [[BCAST_TOPTR]], // CHECK: store [[TT]]* [[D]], [[TT]]** [[BCAST_TOPTR]],
// CHECK: [[PTR_GEP2_8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 8 // CHECK: [[PTR_GEP2_8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 8
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_8]] to [[TT]]** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_8]] to [[TT]]**
// CHECK: store [[TT]]* [[D]], [[TT]]** [[BCAST_TOPTR]], // CHECK: store [[TT]]* [[D]], [[TT]]** [[BCAST_TOPTR]],
// CHECK: [[SIZE_GEPCN_8:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 8
// CHECK: store i{{[0-9]+}} {{[0-9]+}}, i{{[0-9]+}}* [[SIZE_GEPCN_8]],
// CHECK: [[BASE_PTR_GEP_ARG2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 // CHECK: [[BASE_PTR_GEP_ARG2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: [[PTR_GEP_ARG2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 // CHECK: [[PTR_GEP_ARG2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: [[SIZES_ARG2:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 // CHECK: {{.+}} = call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, {{.+}}, i32 9, i8** [[BASE_PTR_GEP_ARG2]], i8** [[PTR_GEP_ARG2]], i[[SZ]]* getelementptr inbounds ([9 x i64], [9 x i64]* [[SIZET2]], i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[MAPT2]], i32 0, i32 0), i8** null, i8** null)
// CHECK: {{.+}} = call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, {{.+}}, i32 9, i8** [[BASE_PTR_GEP_ARG2]], i8** [[PTR_GEP_ARG2]], i[[SZ]]* [[SIZES_ARG2]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[MAPT2]], i32 0, i32 0), i8** null, i8** null)
// make sure that firstprivate variables are generated in all cases and that we use those instances for operations inside the // make sure that firstprivate variables are generated in all cases and that we use those instances for operations inside the
// target region // target region
// TCHECK: define {{.*}}void @__omp_offloading_{{.+}}(i{{[0-9]+}} [[A2_IN:%.+]], [10 x float]* {{.+}} [[B_IN:%.+]], i{{[0-9]+}} [[BN_SZ:%.+]], float* {{.+}} [[BN_IN:%.+]], [5 x [10 x double]]* {{.+}} [[C_IN:%.+]], i{{[0-9]+}} [[CN_SZ1:%.+]], i{{[0-9]+}} [[CN_SZ2:%.+]], double* {{.+}} [[CN_IN:%.+]], [[TT]]* {{.+}} [[D_IN:%.+]]) // TCHECK: define {{.*}}void @__omp_offloading_{{.+}}(i{{[0-9]+}} [[A2_IN:%.+]], [10 x float]* {{.+}} [[B_IN:%.+]], i{{[0-9]+}} [[BN_SZ:%.+]], float* {{.+}} [[BN_IN:%.+]], [5 x [10 x double]]* {{.+}} [[C_IN:%.+]], i{{[0-9]+}} [[CN_SZ1:%.+]], i{{[0-9]+}} [[CN_SZ2:%.+]], double* {{.+}} [[CN_IN:%.+]], [[TT]]* {{.+}} [[D_IN:%.+]])
// TCHECK: [[A2_ADDR:%.+]] = alloca i{{[0-9]+}}, // TCHECK: [[A2_ADDR:%.+]] = alloca i{{[0-9]+}},
// TCHECK: [[B_ADDR:%.+]] = alloca [10 x float]*, // TCHECK: [[B_ADDR:%.+]] = alloca [10 x float]*,
// TCHECK: [[VLA_ADDR:%.+]] = alloca i{{[0-9]+}}, // TCHECK: [[VLA_ADDR:%.+]] = alloca i{{[0-9]+}},
// TCHECK: [[BN_ADDR:%.+]] = alloca float*, // TCHECK: [[BN_ADDR:%.+]] = alloca float*,
▲ Show 20 Lines • Show All 166 Lines • ▼ Show 20 Lines#pragma omp target firstprivate(b, c)
return c[1][1] + (int)b; return c[1][1] + (int)b;
} }
// on the host side, we first generate r1, then the static function and the template above // on the host side, we first generate r1, then the static function and the template above
// CHECK: define{{.+}} i32 {{.+}}([[S1]]* {{.+}}, i{{[0-9]+}} {{.+}}) // CHECK: define{{.+}} i32 {{.+}}([[S1]]* {{.+}}, i{{[0-9]+}} {{.+}})
// CHECK: [[BASE_PTRS4:%.+]] = alloca [5 x i8*], // CHECK: [[BASE_PTRS4:%.+]] = alloca [5 x i8*],
// CHECK: [[PTRS4:%.+]] = alloca [5 x i8*], // CHECK: [[PTRS4:%.+]] = alloca [5 x i8*],
// CHECK: [[SIZET4:%.+]] = alloca [5 x i{{[0-9]+}}],
// map(this: this ptr is implicitly captured (not firstprivate matter) // map(this: this ptr is implicitly captured (not firstprivate matter)
// CHECK: [[BP0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 // CHECK: [[BP0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: [[CBP0:%.+]] = bitcast i8** [[BP0]] to %struct.S1** // CHECK: [[CBP0:%.+]] = bitcast i8** [[BP0]] to %struct.S1**
// CHECK: store %struct.S1* [[THIS:%.+]], %struct.S1** [[CBP0]], // CHECK: store %struct.S1* [[THIS:%.+]], %struct.S1** [[CBP0]],
// CHECK: [[P0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 // CHECK: [[P0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: [[CP0:%.+]] = bitcast i8** [[P0]] to double** // CHECK: [[CP0:%.+]] = bitcast i8** [[P0]] to double**
// CHECK: store double* [[A:%.+]], double** [[CP0]], // CHECK: store double* [[A:%.+]], double** [[CP0]],
// CHECK: [[SZ0:%.+]] = getelementptr inbounds [5 x i{{[0-9]+}}], [5 x i{{[0-9]+}}]* [[SIZET4]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CHECK: store i{{[0-9]+}} 8, i{{[0-9]+}}* [[SZ0]],
// firstprivate(b): base_ptr = b, ptr = b, size = 4 (pass by-value) // firstprivate(b): base_ptr = b, ptr = b, size = 4 (pass by-value)
// CHECK: [[BASE_PTRS_GEP4_1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 // CHECK: [[BASE_PTRS_GEP4_1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_1]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_1]] to i{{[0-9]+}}*
// CHECK: store i{{[0-9]+}} [[B_CAST:%.+]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK: store i{{[0-9]+}} [[B_CAST:%.+]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[PTRS_GEP4_1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 // CHECK: [[PTRS_GEP4_1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_1]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_1]] to i{{[0-9]+}}*
// CHECK: store i{{[0-9]+}} [[B_CAST]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK: store i{{[0-9]+}} [[B_CAST]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[SIZES_GEP4_1:%.+]] = getelementptr inbounds [5 x i{{[0-9]+}}], [5 x i{{[0-9]+}}]* [[SIZET4]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
// CHECK: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIZES_GEP4_1]],
// firstprivate(c), 3 entries: 2, n, c // firstprivate(c), 3 entries: 2, n, c
// CHECK: [[BASE_PTRS_GEP4_2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 // CHECK: [[BASE_PTRS_GEP4_2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_2]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_2]] to i{{[0-9]+}}*
// CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[PTRS_GEP4_2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 // CHECK: [[PTRS_GEP4_2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_2]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_2]] to i{{[0-9]+}}*
// CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[SIZES_GEP4_2:%.+]] = getelementptr inbounds [5 x i{{[0-9]+}}], [5 x i{{[0-9]+}}]* [[SIZET4]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
// CHECK-64: store i{{[0-9]+}} 8, i{{[0-9]+}}* [[SIZES_GEP4_2]],
// CHECK-32: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIZES_GEP4_2]],
// CHECK: [[BASE_PTRS_GEP4_3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 3 // CHECK: [[BASE_PTRS_GEP4_3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_3]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_3]] to i{{[0-9]+}}*
// CHECK: store i{{[0-9]+}} [[N:%.+]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK: store i{{[0-9]+}} [[N:%.+]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[PTRS_GEP4_3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 3 // CHECK: [[PTRS_GEP4_3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_3]] to i{{[0-9]+}}* // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_3]] to i{{[0-9]+}}*
// CHECK: store i{{[0-9]+}} [[N]], i{{[0-9]+}}* [[BCAST_TOPTR]], // CHECK: store i{{[0-9]+}} [[N]], i{{[0-9]+}}* [[BCAST_TOPTR]],
// CHECK: [[SIZES_GEP4_3:%.+]] = getelementptr inbounds [5 x i{{[0-9]+}}], [5 x i{{[0-9]+}}]* [[SIZET4]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
// CHECK-64: store i{{[0-9]+}} 8, i{{[0-9]+}}* [[SIZES_GEP4_3]],
// CHECK-32: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIZES_GEP4_3]],
// CHECK: [[BASE_PTRS_GEP4_4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 4 // CHECK: [[BASE_PTRS_GEP4_4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_4]] to i{{[0-9]+}}** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_4]] to i{{[0-9]+}}**
// CHECK: store i{{[0-9]+}}* [[B:%.+]], i{{[0-9]+}}** [[BCAST_TOPTR]], // CHECK: store i{{[0-9]+}}* [[B:%.+]], i{{[0-9]+}}** [[BCAST_TOPTR]],
// CHECK: [[PTRS_GEP4_4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 4 // CHECK: [[PTRS_GEP4_4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
// CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_4]] to i{{[0-9]+}}** // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_4]] to i{{[0-9]+}}**
// CHECK: store i{{[0-9]+}}* [[B]], i{{[0-9]+}}** [[BCAST_TOPTR]], // CHECK: store i{{[0-9]+}}* [[B]], i{{[0-9]+}}** [[BCAST_TOPTR]],
// CHECK: [[SIZES_GEP4_4:%.+]] = getelementptr inbounds [5 x i{{[0-9]+}}], [5 x i{{[0-9]+}}]* [[SIZET4]], i{{[0-9]+}} 0, i{{[0-9]+}} 4 // CHECK: store i{{[0-9]+}} [[B_SIZE:%.+]], i{{[0-9]+}}* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZET4]], i32 0, i32 4),
// CHECK: store i{{[0-9]+}} [[B_SIZE:%.+]], i{{[0-9]+}}* [[SIZES_GEP4_4]],
// only check that we use the map types stored in the global variable // only check that we use the map types stored in the global variable
// CHECK: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, {{.+}}, i32 5, i8** {{.+}}, i8** {{.+}}, i{{[0-9]+}}* {{.+}}, i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPT4]], i32 0, i32 0), i8** null, i8** null) // CHECK: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, {{.+}}, i32 5, i8** {{.+}}, i8** {{.+}}, i{{[0-9]+}}* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZET4]], i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPT4]], i32 0, i32 0), i8** null, i8** null)
// TCHECK: define weak void @__omp_offloading_{{.+}}([[S1]]* [[TH:%.+]], i{{[0-9]+}} [[B_IN:%.+]], i{{[0-9]+}} [[VLA:%.+]], i{{[0-9]+}} [[VLA1:%.+]], i{{[0-9]+}}{{.+}} [[C_IN:%.+]]) // TCHECK: define weak void @__omp_offloading_{{.+}}([[S1]]* [[TH:%.+]], i{{[0-9]+}} [[B_IN:%.+]], i{{[0-9]+}} [[VLA:%.+]], i{{[0-9]+}} [[VLA1:%.+]], i{{[0-9]+}}{{.+}} [[C_IN:%.+]])
// TCHECK: [[TH_ADDR:%.+]] = alloca [[S1]]*, // TCHECK: [[TH_ADDR:%.+]] = alloca [[S1]]*,
// TCHECK: [[B_ADDR:%.+]] = alloca i{{[0-9]+}}, // TCHECK: [[B_ADDR:%.+]] = alloca i{{[0-9]+}},
// TCHECK: [[VLA_ADDR:%.+]] = alloca i{{[0-9]+}}, // TCHECK: [[VLA_ADDR:%.+]] = alloca i{{[0-9]+}},
// TCHECK: [[VLA_ADDR2:%.+]] = alloca i{{[0-9]+}}, // TCHECK: [[VLA_ADDR2:%.+]] = alloca i{{[0-9]+}},
// TCHECK: [[C_ADDR:%.+]] = alloca i{{[0-9]+}}*, // TCHECK: [[C_ADDR:%.+]] = alloca i{{[0-9]+}}*,
// TCHECK-NOT: alloca i{{[0-9]+}}, // TCHECK-NOT: alloca i{{[0-9]+}},
▲ Show 20 Lines • Show All 122 Lines • Show Last 20 Lines

clang/test/OpenMP/target_map_codegen_12.cpp

Show All 29 Lines
// RUN: %clang_cc1 -DCK13 -verify -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY12 %s // RUN: %clang_cc1 -DCK13 -verify -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY12 %s
// RUN: %clang_cc1 -DCK13 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK13 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY12 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY12 %s
// SIMD-ONLY12-NOT: {{__kmpc|__tgt}} // SIMD-ONLY12-NOT: {{__kmpc|__tgt}}
#ifdef CK13 #ifdef CK13
// CK13-LABEL: @.__omp_offloading_{{.*}}implicit_maps_variable_length_array{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK13-LABEL: @.__omp_offloading_{{.*}}implicit_maps_variable_length_array{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// We don't have a constant map size for VLAs. // CK13-DAG: [[SIZES:@.+]] = {{.+}}global [3 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 0]
// Map types: // Map types:
// - OMP_MAP_PRIVATE_VAL + OMP_MAP_TARGET_PARAM + OMP_MAP_IMPLICIT = 800 (vla size) // - OMP_MAP_PRIVATE_VAL + OMP_MAP_TARGET_PARAM + OMP_MAP_IMPLICIT = 800 (vla size)
// - OMP_MAP_PRIVATE_VAL + OMP_MAP_TARGET_PARAM + OMP_MAP_IMPLICIT = 800 (vla size) // - OMP_MAP_PRIVATE_VAL + OMP_MAP_TARGET_PARAM + OMP_MAP_IMPLICIT = 800 (vla size)
// - OMP_MAP_TO + OMP_MAP_FROM + OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 547 // - OMP_MAP_TO + OMP_MAP_FROM + OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 547
// CK13-DAG: [[TYPES:@.+]] = {{.+}}constant [3 x i64] [i64 800, i64 800, i64 547] // CK13-DAG: [[TYPES:@.+]] = {{.+}}constant [3 x i64] [i64 800, i64 800, i64 547]
// CK13-LABEL: implicit_maps_variable_length_array{{.*}}( // CK13-LABEL: implicit_maps_variable_length_array{{.*}}(
void implicit_maps_variable_length_array (int a){ void implicit_maps_variable_length_array (int a){
double vla[2][a]; double vla[2][a];
// CK13-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{.+}}, i8* {{.+}}, i32 3, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], i64* [[SGEP:%[^,]+]], {{.+}}[[TYPES]]{{.+}}, i8** null, i8** null) // CK13-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{.+}}, i8* {{.+}}, i32 3, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}}, i8** null, i8** null)
// CK13-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0 // CK13-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
// CK13-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0 // CK13-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
// CK13-DAG: [[SGEP]] = getelementptr inbounds {{.+}}[[SS:%[^,]+]], i32 0, i32 0
// CK13-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0 // CK13-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
// CK13-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0 // CK13-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
// CK13-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[SS]], i32 0, i32 0
// CK13-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[sz:64|32]]* // CK13-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[sz:64|32]]*
// CK13-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[sz]]* // CK13-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[sz]]*
// CK13-DAG: store i[[sz]] 2, i[[sz]]* [[CBP0]] // CK13-DAG: store i[[sz]] 2, i[[sz]]* [[CBP0]]
// CK13-DAG: store i[[sz]] 2, i[[sz]]* [[CP0]] // CK13-DAG: store i[[sz]] 2, i[[sz]]* [[CP0]]
// CK13-DAG: store i64 {{8|4}}, i64* [[S0]],
// CK13-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1 // CK13-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1
// CK13-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1 // CK13-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1
// CK13-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[SS]], i32 0, i32 1
// CK13-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[sz]]* // CK13-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[sz]]*
// CK13-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[sz]]* // CK13-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[sz]]*
// CK13-DAG: store i[[sz]] [[VAL:%.+]], i[[sz]]* [[CBP1]] // CK13-DAG: store i[[sz]] [[VAL:%.+]], i[[sz]]* [[CBP1]]
// CK13-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP1]] // CK13-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP1]]
// CK13-DAG: store i64 {{8|4}}, i64* [[S1]],
// CK13-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2 // CK13-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2
// CK13-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2 // CK13-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2
// CK13-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[SS]], i32 0, i32 2
// CK13-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to double** // CK13-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to double**
// CK13-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double** // CK13-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double**
// CK13-DAG: store double* [[DECL:%.+]], double** [[CBP2]] // CK13-DAG: store double* [[DECL:%.+]], double** [[CBP2]]
// CK13-DAG: store double* [[DECL]], double** [[CP2]] // CK13-DAG: store double* [[DECL]], double** [[CP2]]
// CK13-DAG: store i64 [[VALS2:%.+]], i64* [[S2]], // CK13-DAG: store i64 [[VALS2:%.+]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZES]], i32 0, i32 2),
// CK13-DAG: [[VALS2]] = {{mul nuw i64 %.+, 8|sext i32 %.+ to i64}} // CK13-DAG: [[VALS2]] = {{mul nuw i64 %.+, 8|sext i32 %.+ to i64}}
// CK13: call void [[KERNEL:@.+]](i[[sz]] {{.+}}, i[[sz]] {{.+}}, double* [[DECL]]) // CK13: call void [[KERNEL:@.+]](i[[sz]] {{.+}}, i[[sz]] {{.+}}, double* [[DECL]])
#pragma omp target #pragma omp target
{ {
vla[1][3] += 1.0; vla[1][3] += 1.0;
} }
} }
Show All 14 Lines

clang/test/OpenMP/target_map_codegen_13.cpp

Show All 32 Lines
// SIMD-ONLY13-NOT: {{__kmpc|__tgt}} // SIMD-ONLY13-NOT: {{__kmpc|__tgt}}
#ifdef CK14 #ifdef CK14
// CK14-DAG: [[ST:%.+]] = type { i32, double } // CK14-DAG: [[ST:%.+]] = type { i32, double }
// CK14-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK14-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK14-DAG: [[SIZES:@.+]] = {{.+}}global [4 x i64] [i64 0, i64 4, i64 8, i64 4]
// Map types: // Map types:
// - OMP_MAP_TARGET_PARAM = 32 // - OMP_MAP_TARGET_PARAM = 32
// - OMP_MAP_TO + OMP_MAP_FROM | OMP_MAP_IMPLICIT | OMP_MAP_MEMBER_OF = 281474976711171 // - OMP_MAP_TO + OMP_MAP_FROM | OMP_MAP_IMPLICIT | OMP_MAP_MEMBER_OF = 281474976711171
// - OMP_MAP_PRIVATE_VAL + OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 800 // - OMP_MAP_PRIVATE_VAL + OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 800
// CK14-DAG: [[TYPES:@.+]] = {{.+}}constant [4 x i64] [i64 32, i64 281474976711171, i64 281474976711171, i64 800] // CK14-DAG: [[TYPES:@.+]] = {{.+}}constant [4 x i64] [i64 32, i64 281474976711171, i64 281474976711171, i64 800]
class SSS { class SSS {
public: public:
Show All 11 Linespublic:
SSS(int a, double b) : a(a), b(b) {} SSS(int a, double b) : a(a), b(b) {}
}; };
// CK14-LABEL: implicit_maps_class{{.*}}( // CK14-LABEL: implicit_maps_class{{.*}}(
void implicit_maps_class (int a){ void implicit_maps_class (int a){
SSS sss(a, (double)a); SSS sss(a, (double)a);
// CK14: define {{.*}}void @{{.+}}foo{{.+}}([[ST]]* {{[^,]+}}, i32 {{[^,]+}}) // CK14: define {{.*}}void @{{.+}}foo{{.+}}([[ST]]* {{[^,]+}}, i32 {{[^,]+}})
// CK14-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{.+}}, i8* {{.+}}, i32 4, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], i64* [[SIZES:%[^,]+]], {{.+}}[[TYPES]]{{.+}}, i8** null, i8** null) // CK14-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{.+}}, i8* {{.+}}, i32 4, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}}, i8** null, i8** null)
// CK14-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0 // CK14-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
// CK14-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0 // CK14-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
// CK14-DAG: [[SIZES]] = getelementptr inbounds {{.+}}[[S:%[^,]+]], i32 0, i32 0
// CK14-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0 // CK14-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
// CK14-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0 // CK14-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
// CK14-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 0
// CK14-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK14-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK14-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK14-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK14-DAG: store [[ST]]* [[DECL:%.+]], [[ST]]** [[CBP0]] // CK14-DAG: store [[ST]]* [[DECL:%.+]], [[ST]]** [[CBP0]]
// CK14-DAG: store i32* [[A:%.+]], i32** [[CP0]] // CK14-DAG: store i32* [[A:%.+]], i32** [[CP0]]
// CK14-DAG: store i64 %{{.+}}, i64* [[S0]] // CK14-DAG: store i64 %{{.+}}, i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZES]], i32 0, i32 0),
// CK14-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1 // CK14-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1
// CK14-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1 // CK14-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1
// CK14-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 1
// CK14-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]** // CK14-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]**
// CK14-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK14-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK14-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP1]] // CK14-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP1]]
// CK14-DAG: store i32* [[A]], i32** [[CP1]] // CK14-DAG: store i32* [[A]], i32** [[CP1]]
// CK14-DAG: store i64 4, i64* [[S1]]
// CK14-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2 // CK14-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2
// CK14-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2 // CK14-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2
// CK14-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 2
// CK14-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]** // CK14-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]**
// CK14-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double** // CK14-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double**
// CK14-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP2]] // CK14-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP2]]
// CK14-DAG: store double* %{{.+}}, double** [[CP2]] // CK14-DAG: store double* %{{.+}}, double** [[CP2]]
// CK14-DAG: store i64 8, i64* [[S2]]
// CK14-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 3 // CK14-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 3
// CK14-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 3 // CK14-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 3
// CK14-DAG: [[S3:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 3
// CK14-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to i[[sz:64|32]]* // CK14-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to i[[sz:64|32]]*
// CK14-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to i[[sz]]* // CK14-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to i[[sz]]*
// CK14-DAG: store i[[sz]] [[VAL:%.+]], i[[sz]]* [[CBP3]] // CK14-DAG: store i[[sz]] [[VAL:%.+]], i[[sz]]* [[CBP3]]
// CK14-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP3]] // CK14-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP3]]
// CK14-DAG: store i64 4, i64* [[S3]]
// CK14-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]], // CK14-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
// CK14-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32* // CK14-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
// CK14-64-DAG: store i32 {{.+}}, i32* [[CADDR]], // CK14-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
// CK14: call void [[KERNEL:@.+]]([[ST]]* [[DECL]], i[[sz]] {{.+}}) // CK14: call void [[KERNEL:@.+]]([[ST]]* [[DECL]], i[[sz]] {{.+}})
sss.foo(123); sss.foo(123);
} }
Show All 13 Lines

clang/test/OpenMP/target_map_codegen_14.cpp

Show All 28 Lines
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY14 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY14 %s
// RUN: %clang_cc1 -DCK15 -verify -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY14 %s // RUN: %clang_cc1 -DCK15 -verify -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY14 %s
// RUN: %clang_cc1 -DCK15 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK15 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY14 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY14 %s
// SIMD-ONLY14-NOT: {{__kmpc|__tgt}} // SIMD-ONLY14-NOT: {{__kmpc|__tgt}}
#ifdef CK15 #ifdef CK15
// CK15: [[ST:%.+]] = type { i32, double } // CK15: [[ST:%.+]] = type { i32, double }
// CK15: [[SIZES:@.+]] = {{.+}}global [4 x i64] [i64 0, i64 4, i64 8, i64 4]
// Map types: // Map types:
// - OMP_MAP_TARGET_PARAM = 32 // - OMP_MAP_TARGET_PARAM = 32
// - OMP_MAP_TO + OMP_MAP_FROM | OMP_MAP_IMPLICIT | OMP_MAP_MEMBER_OF = 281474976711171 // - OMP_MAP_TO + OMP_MAP_FROM | OMP_MAP_IMPLICIT | OMP_MAP_MEMBER_OF = 281474976711171
// - OMP_MAP_PRIVATE_VAL + OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 800 // - OMP_MAP_PRIVATE_VAL + OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 800
// CK15: [[TYPES:@.+]] = {{.+}}constant [4 x i64] [i64 32, i64 281474976711171, i64 281474976711171, i64 800] // CK15: [[TYPES:@.+]] = {{.+}}constant [4 x i64] [i64 32, i64 281474976711171, i64 281474976711171, i64 800]
// CK15: [[SIZES2:@.+]] = {{.+}}global [4 x i64] [i64 0, i64 4, i64 8, i64 4]
// Map types: // Map types:
// - OMP_MAP_TARGET_PARAM = 32 // - OMP_MAP_TARGET_PARAM = 32
// - OMP_MAP_TO + OMP_MAP_FROM | OMP_MAP_IMPLICIT | OMP_MAP_MEMBER_OF = 281474976711171 // - OMP_MAP_TO + OMP_MAP_FROM | OMP_MAP_IMPLICIT | OMP_MAP_MEMBER_OF = 281474976711171
// - OMP_MAP_PRIVATE_VAL + OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 800 // - OMP_MAP_PRIVATE_VAL + OMP_MAP_TARGET_PARAM | OMP_MAP_IMPLICIT = 800
// CK15: [[TYPES2:@.+]] = {{.+}}constant [4 x i64] [i64 32, i64 281474976711171, i64 281474976711171, i64 800] // CK15: [[TYPES2:@.+]] = {{.+}}constant [4 x i64] [i64 32, i64 281474976711171, i64 281474976711171, i64 800]
template<int x> template<int x>
class SSST { class SSST {
Show All 20 Linespublic:
SSST(int a, double b) : a(a), b(b) {} SSST(int a, double b) : a(a), b(b) {}
}; };
// CK15-LABEL: implicit_maps_templated_class{{.*}}( // CK15-LABEL: implicit_maps_templated_class{{.*}}(
void implicit_maps_templated_class (int a){ void implicit_maps_templated_class (int a){
SSST<123> ssst(a, (double)a); SSST<123> ssst(a, (double)a);
// CK15: define {{.*}}void @{{.+}}foo{{.+}}([[ST]]* {{[^,]+}}, i32 {{[^,]+}}) // CK15: define {{.*}}void @{{.+}}foo{{.+}}([[ST]]* {{[^,]+}}, i32 {{[^,]+}})
// CK15-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{.+}}, i8* {{.+}}, i32 4, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], i64* [[SIZES:%[^,]+]], {{.+}}[[TYPES]]{{.+}}, i8** null, i8** null) // CK15-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{.+}}, i8* {{.+}}, i32 4, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}}, i8** null, i8** null)
// CK15-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0 // CK15-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
// CK15-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0 // CK15-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
// CK15-DAG: [[SIZES]] = getelementptr inbounds {{.+}}[[S:%[^,]+]], i32 0, i32 0
// CK15-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0 // CK15-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
// CK15-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0 // CK15-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
// CK15-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 0
// CK15-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK15-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK15-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK15-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK15-DAG: store [[ST]]* [[DECL:%.+]], [[ST]]** [[CBP0]] // CK15-DAG: store [[ST]]* [[DECL:%.+]], [[ST]]** [[CBP0]]
// CK15-DAG: store i32* [[A:%.+]], i32** [[CP0]] // CK15-DAG: store i32* [[A:%.+]], i32** [[CP0]]
// CK15-DAG: store i64 %{{.+}}, i64* [[S0]] // CK15-DAG: store i64 %{{.+}}, i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZES]], i32 0, i32 0),
// CK15-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1 // CK15-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1
// CK15-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1 // CK15-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1
// CK15-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 1
// CK15-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]** // CK15-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]**
// CK15-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK15-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK15-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP1]] // CK15-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP1]]
// CK15-DAG: store i32* [[A]], i32** [[CP1]] // CK15-DAG: store i32* [[A]], i32** [[CP1]]
// CK15-DAG: store i64 4, i64* [[S1]]
// CK15-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2 // CK15-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2
// CK15-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2 // CK15-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2
// CK15-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 2
// CK15-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]** // CK15-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]**
// CK15-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double** // CK15-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double**
// CK15-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP2]] // CK15-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP2]]
// CK15-DAG: store double* %{{.+}}, double** [[CP2]] // CK15-DAG: store double* %{{.+}}, double** [[CP2]]
// CK15-DAG: store i64 8, i64* [[S2]]
// CK15-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 3 // CK15-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 3
// CK15-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 3 // CK15-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 3
// CK15-DAG: [[S3:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 3
// CK15-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to i[[sz:64|32]]* // CK15-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to i[[sz:64|32]]*
// CK15-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to i[[sz]]* // CK15-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to i[[sz]]*
// CK15-DAG: store i[[sz]] [[VAL:%.+]], i[[sz]]* [[CBP3]] // CK15-DAG: store i[[sz]] [[VAL:%.+]], i[[sz]]* [[CBP3]]
// CK15-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP3]] // CK15-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP3]]
// CK15-DAG: store i64 4, i64* [[S3]]
// CK15-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]], // CK15-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
// CK15-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32* // CK15-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
// CK15-64-DAG: store i32 {{.+}}, i32* [[CADDR]], // CK15-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
// CK15: call void [[KERNEL:@.+]]([[ST]]* [[DECL]], i[[sz]] {{.+}}) // CK15: call void [[KERNEL:@.+]]([[ST]]* [[DECL]], i[[sz]] {{.+}})
ssst.foo(456); ssst.foo(456);
// CK15: define {{.*}}void @{{.+}}bar{{.+}}([[ST]]* {{[^,]+}}, i32 {{[^,]+}}) // CK15: define {{.*}}void @{{.+}}bar{{.+}}([[ST]]* {{[^,]+}}, i32 {{[^,]+}})
// CK15-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{.+}}, i8* {{.+}}, i32 4, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], i64* [[SIZES:[^,]+]], {{.+}}[[TYPES2]]{{.+}}, i8** null, i8** null) // CK15-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{.+}}, i8* {{.+}}, i32 4, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES2]]{{.+}}, {{.+}}[[TYPES2]]{{.+}}, i8** null, i8** null)
// CK15-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0 // CK15-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
// CK15-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0 // CK15-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
// CK15-DAG: [[SIZES]] = getelementptr inbounds {{.+}}[[S:%[^,]+]], i32 0, i32 0
// CK15-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0 // CK15-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
// CK15-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0 // CK15-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
// CK15-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 0
// CK15-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK15-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK15-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK15-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK15-DAG: store [[ST]]* [[DECL:%.+]], [[ST]]** [[CBP0]] // CK15-DAG: store [[ST]]* [[DECL:%.+]], [[ST]]** [[CBP0]]
// CK15-DAG: store i32* [[A:%.+]], i32** [[CP0]] // CK15-DAG: store i32* [[A:%.+]], i32** [[CP0]]
// CK15-DAG: store i64 %{{.+}}, i64* [[S0]] // CK15-DAG: store i64 %{{.+}}, i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZES2]], i32 0, i32 0),
// CK15-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1 // CK15-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1
// CK15-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1 // CK15-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1
// CK15-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 1
// CK15-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]** // CK15-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]**
// CK15-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK15-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK15-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP1]] // CK15-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP1]]
// CK15-DAG: store i32* [[A]], i32** [[CP1]] // CK15-DAG: store i32* [[A]], i32** [[CP1]]
// CK15-DAG: store i64 4, i64* [[S1]]
// CK15-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2 // CK15-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2
// CK15-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2 // CK15-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2
// CK15-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 2
// CK15-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]** // CK15-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]**
// CK15-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double** // CK15-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double**
// CK15-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP2]] // CK15-DAG: store [[ST]]* [[DECL]], [[ST]]** [[CBP2]]
// CK15-DAG: store double* %{{.+}}, double** [[CP2]] // CK15-DAG: store double* %{{.+}}, double** [[CP2]]
// CK15-DAG: store i64 8, i64* [[S2]]
// CK15-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 3 // CK15-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 3
// CK15-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 3 // CK15-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 3
// CK15-DAG: [[S3:%.+]] = getelementptr inbounds {{.+}}[[S]], i32 0, i32 3
// CK15-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to i[[sz]]* // CK15-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to i[[sz]]*
// CK15-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to i[[sz]]* // CK15-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to i[[sz]]*
// CK15-DAG: store i[[sz]] [[VAL:%.+]], i[[sz]]* [[CBP3]] // CK15-DAG: store i[[sz]] [[VAL:%.+]], i[[sz]]* [[CBP3]]
// CK15-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP3]] // CK15-DAG: store i[[sz]] [[VAL]], i[[sz]]* [[CP3]]
// CK15-DAG: store i64 4, i64* [[S3]]
// CK15-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]], // CK15-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
// CK15-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32* // CK15-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
// CK15-64-DAG: store i32 {{.+}}, i32* [[CADDR]], // CK15-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
// CK15: call void [[KERNEL2:@.+]]([[ST]]* [[DECL]], i[[sz]] {{.+}}) // CK15: call void [[KERNEL2:@.+]]([[ST]]* [[DECL]], i[[sz]] {{.+}})
ssst.bar<210>(789); ssst.bar<210>(789);
} }
Show All 24 Lines

clang/test/OpenMP/target_map_codegen_18.inc

Show First 20 Lines • Show All 85 Lines • ▼ Show 20 Lines
// CK19-NOUSE: [[MTYPE14:@.+]] = private {{.*}}constant [1 x i64] [i64 1] // CK19-NOUSE: [[MTYPE14:@.+]] = private {{.*}}constant [1 x i64] [i64 1]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19: [[SIZE15:@.+]] = private {{.*}}constant [1 x i64] [i64 4] // CK19: [[SIZE15:@.+]] = private {{.*}}constant [1 x i64] [i64 4]
// CK19-USE: [[MTYPE15:@.+]] = private {{.*}}constant [1 x i64] [i64 34] // CK19-USE: [[MTYPE15:@.+]] = private {{.*}}constant [1 x i64] [i64 34]
// CK19-NOUSE: [[MTYPE15:@.+]] = private {{.*}}constant [1 x i64] [i64 2] // CK19-NOUSE: [[MTYPE15:@.+]] = private {{.*}}constant [1 x i64] [i64 2]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE16:@.+]] = private {{.*}}global [2 x i64] [i64 {{8|4}}, i64 0]
// CK19-USE: [[MTYPE16:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 33] // CK19-USE: [[MTYPE16:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 33]
// CK19-NOUSE: [[MTYPE16:@.+]] = private {{.*}}constant [1 x i64] [i64 1] // CK19-NOUSE: [[MTYPE16:@.+]] = private {{.*}}constant [1 x i64] [i64 1]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE17:@.+]] = private {{.*}}constant [2 x i64] [i64 {{8|4}}, i64 240] // CK19-USE: [[SIZE17:@.+]] = private {{.*}}constant [2 x i64] [i64 {{8|4}}, i64 240]
// CK19-USE: [[MTYPE17:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 34] // CK19-USE: [[MTYPE17:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 34]
// CK19-NOUSE: [[SIZE17:@.+]] = private {{.*}}constant [1 x i64] [i64 240] // CK19-NOUSE: [[SIZE17:@.+]] = private {{.*}}constant [1 x i64] [i64 240]
// CK19-NOUSE: [[MTYPE17:@.+]] = private {{.*}}constant [1 x i64] [i64 2] // CK19-NOUSE: [[MTYPE17:@.+]] = private {{.*}}constant [1 x i64] [i64 2]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE18:@.+]] = private {{.*}}constant [2 x i64] [i64 {{8|4}}, i64 240] // CK19-USE: [[SIZE18:@.+]] = private {{.*}}constant [2 x i64] [i64 {{8|4}}, i64 240]
// CK19-USE: [[MTYPE18:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 35] // CK19-USE: [[MTYPE18:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 35]
// CK19-NOUSE: [[SIZE18:@.+]] = private {{.*}}constant [1 x i64] [i64 240] // CK19-NOUSE: [[SIZE18:@.+]] = private {{.*}}constant [1 x i64] [i64 240]
// CK19-NOUSE: [[MTYPE18:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE18:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE19:@.+]] = private {{.*}}global [2 x i64] [i64 {{8|4}}, i64 0]
// CK19-USE: [[MTYPE19:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 32] // CK19-USE: [[MTYPE19:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 32]
// CK19-NOUSE: [[MTYPE19:@.+]] = private {{.*}}constant [1 x i64] zeroinitializer // CK19-NOUSE: [[MTYPE19:@.+]] = private {{.*}}constant [1 x i64] zeroinitializer
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE20:@.+]] = private {{.*}}constant [2 x i64] [i64 {{8|4}}, i64 4] // CK19-USE: [[SIZE20:@.+]] = private {{.*}}constant [2 x i64] [i64 {{8|4}}, i64 4]
// CK19-USE: [[MTYPE20:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 33] // CK19-USE: [[MTYPE20:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 33]
// CK19-NOUSE: [[SIZE20:@.+]] = private {{.*}}constant [1 x i64] [i64 4] // CK19-NOUSE: [[SIZE20:@.+]] = private {{.*}}constant [1 x i64] [i64 4]
// CK19-NOUSE: [[MTYPE20:@.+]] = private {{.*}}constant [1 x i64] [i64 1] // CK19-NOUSE: [[MTYPE20:@.+]] = private {{.*}}constant [1 x i64] [i64 1]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE21:@.+]] = private {{.*}}global [2 x i64] [i64 {{8|4}}, i64 0]
// CK19-USE: [[MTYPE21:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 35] // CK19-USE: [[MTYPE21:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 35]
// CK19-NOUSE: [[MTYPE21:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE21:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE22:@.+]] = private {{.*}}constant [2 x i64] [i64 {{8|4}}, i64 4] // CK19-USE: [[SIZE22:@.+]] = private {{.*}}constant [2 x i64] [i64 {{8|4}}, i64 4]
// CK19-USE: [[MTYPE22:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 35] // CK19-USE: [[MTYPE22:@.+]] = private {{.*}}constant [2 x i64] [i64 800, i64 35]
// CK19-NOUSE: [[SIZE22:@.+]] = private {{.*}}constant [1 x i64] [i64 4] // CK19-NOUSE: [[SIZE22:@.+]] = private {{.*}}constant [1 x i64] [i64 4]
// CK19-NOUSE: [[MTYPE22:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE22:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
Show All 29 Lines
// CK19-NOUSE: [[MTYPE28:@.+]] = private {{.*}}constant [3 x i64] [i64 3, i64 16, i64 19] // CK19-NOUSE: [[MTYPE28:@.+]] = private {{.*}}constant [3 x i64] [i64 3, i64 16, i64 19]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19: [[SIZE29:@.+]] = private {{.*}}constant [3 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 4] // CK19: [[SIZE29:@.+]] = private {{.*}}constant [3 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 4]
// CK19-USE: [[MTYPE29:@.+]] = private {{.*}}constant [3 x i64] [i64 35, i64 16, i64 19] // CK19-USE: [[MTYPE29:@.+]] = private {{.*}}constant [3 x i64] [i64 35, i64 16, i64 19]
// CK19-NOUSE: [[MTYPE29:@.+]] = private {{.*}}constant [3 x i64] [i64 3, i64 16, i64 19] // CK19-NOUSE: [[MTYPE29:@.+]] = private {{.*}}constant [3 x i64] [i64 3, i64 16, i64 19]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE30:@.+]] = private {{.*}}global [4 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 {{8|4}}, i64 0]
// CK19-USE: [[MTYPE30:@.+]] = private {{.*}}constant [4 x i64] [i64 800, i64 800, i64 800, i64 35] // CK19-USE: [[MTYPE30:@.+]] = private {{.*}}constant [4 x i64] [i64 800, i64 800, i64 800, i64 35]
// CK19-NOUSE: [[MTYPE30:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE30:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE31:@.+]] = private {{.*}}constant [4 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 {{8|4}}, i64 40] // CK19-USE: [[SIZE31:@.+]] = private {{.*}}constant [4 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 {{8|4}}, i64 40]
// CK19-USE: [[MTYPE31:@.+]] = private {{.*}}constant [4 x i64] [i64 800, i64 800, i64 800, i64 35] // CK19-USE: [[MTYPE31:@.+]] = private {{.*}}constant [4 x i64] [i64 800, i64 800, i64 800, i64 35]
// CK19-NOUSE: [[SIZE31:@.+]] = private {{.*}}constant [1 x i64] [i64 40] // CK19-NOUSE: [[SIZE31:@.+]] = private {{.*}}constant [1 x i64] [i64 40]
// CK19-NOUSE: [[MTYPE31:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE31:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
Show All 18 Lines
// CK19-NOUSE: [[MTYPE35:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE35:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19: [[SIZE36:@.+]] = private {{.*}}constant [1 x i64] [i64 208] // CK19: [[SIZE36:@.+]] = private {{.*}}constant [1 x i64] [i64 208]
// CK19-USE: [[MTYPE36:@.+]] = private {{.*}}constant [1 x i64] [i64 35] // CK19-USE: [[MTYPE36:@.+]] = private {{.*}}constant [1 x i64] [i64 35]
// CK19-NOUSE: [[MTYPE36:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE36:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE37:@.+]] = private {{.*}}global [3 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 0]
// CK19-USE: [[MTYPE37:@.+]] = private {{.*}}constant [3 x i64] [i64 800, i64 800, i64 35] // CK19-USE: [[MTYPE37:@.+]] = private {{.*}}constant [3 x i64] [i64 800, i64 800, i64 35]
// CK19-NOUSE: [[MTYPE37:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE37:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE38:@.+]] = private {{.*}}global [3 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 0]
// CK19-USE: [[MTYPE38:@.+]] = private {{.*}}constant [3 x i64] [i64 800, i64 800, i64 35] // CK19-USE: [[MTYPE38:@.+]] = private {{.*}}constant [3 x i64] [i64 800, i64 800, i64 35]
// CK19-NOUSE: [[MTYPE38:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE38:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE39:@.+]] = private {{.*}}global [3 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 0]
// CK19-USE: [[MTYPE39:@.+]] = private {{.*}}constant [3 x i64] [i64 800, i64 800, i64 35] // CK19-USE: [[MTYPE39:@.+]] = private {{.*}}constant [3 x i64] [i64 800, i64 800, i64 35]
// CK19-NOUSE: [[MTYPE39:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE39:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE40:@.+]] = private {{.*}}global [3 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 0]
// CK19-USE: [[MTYPE40:@.+]] = private {{.*}}constant [3 x i64] [i64 800, i64 800, i64 35] // CK19-USE: [[MTYPE40:@.+]] = private {{.*}}constant [3 x i64] [i64 800, i64 800, i64 35]
// CK19-NOUSE: [[MTYPE40:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE40:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK19-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK19-USE: [[SIZE41:@.+]] = private {{.*}}constant [3 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 208] // CK19-USE: [[SIZE41:@.+]] = private {{.*}}constant [3 x i64] [i64 {{8|4}}, i64 {{8|4}}, i64 208]
// CK19-USE: [[MTYPE41:@.+]] = private {{.*}}constant [3 x i64] [i64 800, i64 800, i64 35] // CK19-USE: [[MTYPE41:@.+]] = private {{.*}}constant [3 x i64] [i64 800, i64 800, i64 35]
// CK19-NOUSE: [[SIZE41:@.+]] = private {{.*}}constant [1 x i64] [i64 208] // CK19-NOUSE: [[SIZE41:@.+]] = private {{.*}}constant [1 x i64] [i64 208]
// CK19-NOUSE: [[MTYPE41:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK19-NOUSE: [[MTYPE41:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
▲ Show 20 Lines • Show All 424 Lines • ▼ Show 20 Lines#ifdef USE
pa[15]++; pa[15]++;
#endif #endif
} }
// Map of a variable-size array. // Map of a variable-size array.
int va[ii]; int va[ii];
// Region 16 // Region 16
// CK19-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|2}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[{{1|2}} x i{{.+}}]* [[MTYPE16]]{{.+}}, i8** null, i8** null) // CK19-USE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|2}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE16]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE16]]{{.+}}, i8** null, i8** null)
// CK19-NOUSE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|2}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE16]]{{.+}}, i8** null, i8** null)
// CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK19-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z:64|32]]* // CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z:64|32]]*
// CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]* // CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CBP0]] // CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CBP0]]
// CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CP0]] // CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CP0]]
// CK19-USE-DAG: store i{{.+}} {{8|4}}, i{{.+}}* [[S0]]
// CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i32** // CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i32**
// CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK19-USE-DAG: store i32* [[VAR1:%.+]], i32** [[CBP1]] // CK19-USE-DAG: store i32* [[VAR1:%.+]], i32** [[CBP1]]
// CK19-USE-DAG: store i32* [[VAR1]], i32** [[CP1]] // CK19-USE-DAG: store i32* [[VAR1]], i32** [[CP1]]
// CK19-USE-DAG: store i{{.+}} [[CSVAL1:%[^,]+]], i{{.+}}* [[S1]] // CK19-USE-DAG: store i{{.+}} [[CSVAL1:%[^,]+]], i{{.+}}* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE16]], i32 0, i32 1),
// CK19-USE-DAG: [[CSVAL1]] = {{mul nuw i64 %.*, 4|sext i32 .+ to i64}} // CK19-USE-DAG: [[CSVAL1]] = {{mul nuw i64 %.*, 4|sext i32 .+ to i64}}
// CK19-NOUSE-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i32** // CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i32**
// CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK19-NOUSE-DAG: store i32* [[VAR0:%.+]], i32** [[CBP0]] // CK19-NOUSE-DAG: store i32* [[VAR0:%.+]], i32** [[CBP0]]
// CK19-NOUSE-DAG: store i32* [[VAR0]], i32** [[CP0]] // CK19-NOUSE-DAG: store i32* [[VAR0]], i32** [[CP0]]
// CK19-NOUSE-DAG: store i{{.+}} [[CSVAL0:%[^,]+]], i{{.+}}* [[S0]] // CK19-NOUSE-DAG: store i{{.+}} [[CSVAL0:%[^,]+]], i{{.+}}* [[S0]]
▲ Show 20 Lines • Show All 78 Lines • ▼ Show 20 Lines#endif
#pragma omp target map(tofrom:va[:60]) #pragma omp target map(tofrom:va[:60])
{ {
#ifdef USE #ifdef USE
va[50]++; va[50]++;
#endif #endif
} }
// Region 19 // Region 19
// CK19-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|2}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[{{1|2}} x i{{.+}}]* [[MTYPE19]]{{.+}}, i8** null, i8** null) // CK19-USE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|2}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE19]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE19]]{{.+}}, i8** null, i8** null)
// CK19-NOUSE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|2}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE19]]{{.+}}, i8** null, i8** null)
// CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK19-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]* // CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]*
// CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]* // CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CBP0]] // CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CBP0]]
// CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CP0]] // CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CP0]]
// CK19-USE-DAG: store i{{.+}} {{8|4}}, i{{.+}}* [[S0]]
// CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i32** // CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i32**
// CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK19-USE-DAG: store i32* [[VAR1:%.+]], i32** [[CBP1]] // CK19-USE-DAG: store i32* [[VAR1:%.+]], i32** [[CBP1]]
// CK19-USE-DAG: store i32* [[SEC1:%.+]], i32** [[CP1]] // CK19-USE-DAG: store i32* [[SEC1:%.+]], i32** [[CP1]]
// CK19-USE-DAG: store i{{.+}} [[CSVAL1:%[^,]+]], i{{.+}}* [[S1]] // CK19-USE-DAG: store i{{.+}} [[CSVAL1:%[^,]+]], i{{.+}}* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE19]], i32 0, i32 1),
// CK19-USE-DAG: [[CSVAL1]] = {{mul nuw i64 %.*, 4|sext i32 .+ to i64}} // CK19-USE-DAG: [[CSVAL1]] = {{mul nuw i64 %.*, 4|sext i32 .+ to i64}}
// CK19-USE-DAG: [[SEC1]] = getelementptr {{.*}}i32* [[VAR1]], i{{.+}} 0 // CK19-USE-DAG: [[SEC1]] = getelementptr {{.*}}i32* [[VAR1]], i{{.+}} 0
// CK19-NOUSE-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i32** // CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i32**
// CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK19-NOUSE-DAG: store i32* [[VAR0:%.+]], i32** [[CBP0]] // CK19-NOUSE-DAG: store i32* [[VAR0:%.+]], i32** [[CBP0]]
// CK19-NOUSE-DAG: store i32* [[SEC0:%.+]], i32** [[CP0]] // CK19-NOUSE-DAG: store i32* [[SEC0:%.+]], i32** [[CP0]]
// CK19-NOUSE-DAG: store i{{.+}} [[CSVAL0:%[^,]+]], i{{.+}}* [[S0]] // CK19-NOUSE-DAG: store i{{.+}} [[CSVAL0:%[^,]+]], i{{.+}}* [[S0]]
▲ Show 20 Lines • Show All 42 Lines • ▼ Show 20 Lines#endif
#pragma omp target map(to:va[15]) #pragma omp target map(to:va[15])
{ {
#ifdef USE #ifdef USE
va[15]++; va[15]++;
#endif #endif
} }
// Region 21 // Region 21
// CK19-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|2}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[{{1|2}} x i{{.+}}]* [[MTYPE21]]{{.+}}, i8** null, i8** null) // CK19-USE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|2}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE21]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE21]]{{.+}}, i8** null, i8** null)
// CK19-NOUSE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|2}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE21]]{{.+}}, i8** null, i8** null)
// CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK19-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]* // CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]*
// CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]* // CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CBP0]] // CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CBP0]]
// CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CP0]] // CK19-USE-DAG: store i[[Z]] {{%.+}}, i[[Z]]* [[CP0]]
// CK19-USE-DAG: store i{{.+}} {{8|4}}, i{{.+}}* [[S0]]
// CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i32** // CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i32**
// CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK19-USE-DAG: store i32* [[VAR1:%.+]], i32** [[CBP1]] // CK19-USE-DAG: store i32* [[VAR1:%.+]], i32** [[CBP1]]
// CK19-USE-DAG: store i32* [[SEC1:%.+]], i32** [[CP1]] // CK19-USE-DAG: store i32* [[SEC1:%.+]], i32** [[CP1]]
// CK19-USE-DAG: store i{{.+}} [[CSVAL1:%[^,]+]], i{{.+}}* [[S1]] // CK19-USE-DAG: store i{{.+}} [[CSVAL1:%[^,]+]], i{{.+}}* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE21]], i32 0, i32 1),
// CK19-USE-DAG: [[CSVAL1]] = {{mul nuw i64 %.*, 4|sext i32 .+ to i64}} // CK19-USE-DAG: [[CSVAL1]] = {{mul nuw i64 %.*, 4|sext i32 .+ to i64}}
// CK19-USE-DAG: [[SEC1]] = getelementptr {{.*}}i32* [[VAR1]], i{{.+}} %{{.+}} // CK19-USE-DAG: [[SEC1]] = getelementptr {{.*}}i32* [[VAR1]], i{{.+}} %{{.+}}
// CK19-NOUSE-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i32** // CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i32**
// CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK19-NOUSE-DAG: store i32* [[VAR0:%.+]], i32** [[CBP0]] // CK19-NOUSE-DAG: store i32* [[VAR0:%.+]], i32** [[CBP0]]
// CK19-NOUSE-DAG: store i32* [[SEC0:%.+]], i32** [[CP0]] // CK19-NOUSE-DAG: store i32* [[SEC0:%.+]], i32** [[CP0]]
// CK19-NOUSE-DAG: store i{{.+}} [[CSVAL0:%[^,]+]], i{{.+}}* [[S0]] // CK19-NOUSE-DAG: store i{{.+}} [[CSVAL0:%[^,]+]], i{{.+}}* [[S0]]
▲ Show 20 Lines • Show All 260 Lines • ▼ Show 20 Lines#ifdef USE
mptr[1][2][3]++; mptr[1][2][3]++;
#endif #endif
} }
// Multidimensional VLA. // Multidimensional VLA.
double mva[23][ii][ii+5]; double mva[23][ii][ii+5];
// Region 30 // Region 30
// CK19-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|4}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[{{1|4}} x i{{.+}}]* [[MTYPE30]]{{.+}}, i8** null, i8** null) // CK19-USE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|4}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[SIZE30]]{{.+}}, {{.+}}getelementptr {{.+}}[{{1|4}} x i{{.+}}]* [[MTYPE30]]{{.+}}, i8** null, i8** null)
// CK19-NOUSE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|4}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE30]]{{.+}}, i8** null, i8** null)
// CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK19-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// //
// CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]* // CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]*
// CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]* // CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] 23, i[[Z]]* [[CBP0]] // CK19-USE-DAG: store i[[Z]] 23, i[[Z]]* [[CBP0]]
// CK19-USE-DAG: store i[[Z]] 23, i[[Z]]* [[CP0]] // CK19-USE-DAG: store i[[Z]] 23, i[[Z]]* [[CP0]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S0]]
// //
// CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[Z]]* // CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[Z]]*
// CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[Z]]* // CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] [[VAR1:%.+]], i[[Z]]* [[CBP1]] // CK19-USE-DAG: store i[[Z]] [[VAR1:%.+]], i[[Z]]* [[CBP1]]
// CK19-USE-DAG: store i[[Z]] [[VAR11:%.+]], i[[Z]]* [[CP1]] // CK19-USE-DAG: store i[[Z]] [[VAR11:%.+]], i[[Z]]* [[CP1]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S1]]
// CK19-64-USE-DAG: [[VAR1]] = zext i32 %{{[^,]+}} to i64 // CK19-64-USE-DAG: [[VAR1]] = zext i32 %{{[^,]+}} to i64
// CK19-64-USE-DAG: [[VAR11]] = zext i32 %{{[^,]+}} to i64 // CK19-64-USE-DAG: [[VAR11]] = zext i32 %{{[^,]+}} to i64
// //
// CK19-USE-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK19-USE-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK19-USE-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to i[[Z]]* // CK19-USE-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to i[[Z]]*
// CK19-USE-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i[[Z]]* // CK19-USE-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] [[VAR2:%.+]], i[[Z]]* [[CBP2]] // CK19-USE-DAG: store i[[Z]] [[VAR2:%.+]], i[[Z]]* [[CBP2]]
// CK19-USE-DAG: store i[[Z]] [[VAR22:%.+]], i[[Z]]* [[CP2]] // CK19-USE-DAG: store i[[Z]] [[VAR22:%.+]], i[[Z]]* [[CP2]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S2]]
// CK19-64-USE-DAG: [[VAR2]] = zext i32 %{{[^,]+}} to i64 // CK19-64-USE-DAG: [[VAR2]] = zext i32 %{{[^,]+}} to i64
// CK19-64-USE-DAG: [[VAR22]] = zext i32 %{{[^,]+}} to i64 // CK19-64-USE-DAG: [[VAR22]] = zext i32 %{{[^,]+}} to i64
// //
// CK19-USE-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3 // CK19-USE-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3
// CK19-USE-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3 // CK19-USE-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3
// CK19-USE-DAG: [[S3:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 3
// CK19-USE-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to double** // CK19-USE-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to double**
// CK19-USE-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to double** // CK19-USE-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to double**
// CK19-USE-DAG: store double* [[VAR3:%.+]], double** [[CBP3]] // CK19-USE-DAG: store double* [[VAR3:%.+]], double** [[CBP3]]
// CK19-USE-DAG: store double* [[VAR3]], double** [[CP3]] // CK19-USE-DAG: store double* [[VAR3]], double** [[CP3]]
// CK19-USE-DAG: store i64 [[CSVAL3:%[^,]+]], i64* [[S3]] // CK19-USE-DAG: store i64 [[CSVAL3:%[^,]+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE30]], i32 0, i32 3),
// CK19-USE-DAG: [[CSVAL3]] = {{mul nuw i64 %[^,]+, 8|sext i32 .+ to i64}} // CK19-USE-DAG: [[CSVAL3]] = {{mul nuw i64 %[^,]+, 8|sext i32 .+ to i64}}
// CK19-NOUSE-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to double** // CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to double**
// CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double** // CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double**
// CK19-NOUSE-DAG: store double* [[VAR0:%.+]], double** [[CBP0]] // CK19-NOUSE-DAG: store double* [[VAR0:%.+]], double** [[CBP0]]
// CK19-NOUSE-DAG: store double* [[VAR0]], double** [[CP0]] // CK19-NOUSE-DAG: store double* [[VAR0]], double** [[CP0]]
// CK19-NOUSE-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]] // CK19-NOUSE-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]]
▲ Show 20 Lines • Show All 185 Lines • ▼ Show 20 Lines#endif
#pragma omp target map(marras[:1][:2][:13]) #pragma omp target map(marras[:1][:2][:13])
{ {
#ifdef USE #ifdef USE
marras[1][2][3]++; marras[1][2][3]++;
#endif #endif
} }
// Region 37 // Region 37
// CK19-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[{{1|3}} x i{{.+}}]* [[MTYPE37]]{{.+}}, i8** null, i8** null) // CK19-USE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[SIZE37]]{{.+}}, {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE37]]{{.+}}, i8** null, i8** null)
// CK19-NOUSE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE37]]{{.+}}, i8** null, i8** null)
// CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK19-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// //
// CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]* // CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]*
// CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]* // CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CBP0]] // CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CBP0]]
// CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CP0]] // CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CP0]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S0]]
// //
// CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[Z]]* // CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[Z]]*
// CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[Z]]* // CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] [[VAR1:%.+]], i[[Z]]* [[CBP1]] // CK19-USE-DAG: store i[[Z]] [[VAR1:%.+]], i[[Z]]* [[CBP1]]
// CK19-USE-DAG: store i[[Z]] [[VAR11:%.+]], i[[Z]]* [[CP1]] // CK19-USE-DAG: store i[[Z]] [[VAR11:%.+]], i[[Z]]* [[CP1]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S1]]
// //
// CK19-USE-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK19-USE-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK19-USE-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [13 x double]** // CK19-USE-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [13 x double]**
// CK19-USE-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to [13 x double]** // CK19-USE-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to [13 x double]**
// CK19-USE-DAG: store [13 x double]* [[VAR2:%.+]], [13 x double]** [[CBP2]] // CK19-USE-DAG: store [13 x double]* [[VAR2:%.+]], [13 x double]** [[CBP2]]
// CK19-USE-DAG: store [13 x double]* [[VAR2]], [13 x double]** [[CP2]] // CK19-USE-DAG: store [13 x double]* [[VAR2]], [13 x double]** [[CP2]]
// CK19-USE-DAG: store i64 [[CSVAL2:%[^,]+]], i64* [[S2]] // CK19-USE-DAG: store i64 [[CSVAL2:%[^,]+]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE37]], i32 0, i32 2),
// CK19-USE-DAG: [[CSVAL2]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}} // CK19-USE-DAG: [[CSVAL2]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}}
// CK19-NOUSE-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [13 x double]** // CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [13 x double]**
// CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [13 x double]** // CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [13 x double]**
// CK19-NOUSE-DAG: store [13 x double]* [[VAR0:%.+]], [13 x double]** [[CBP0]] // CK19-NOUSE-DAG: store [13 x double]* [[VAR0:%.+]], [13 x double]** [[CBP0]]
// CK19-NOUSE-DAG: store [13 x double]* [[VAR0]], [13 x double]** [[CP0]] // CK19-NOUSE-DAG: store [13 x double]* [[VAR0]], [13 x double]** [[CP0]]
// CK19-NOUSE-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]] // CK19-NOUSE-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]]
// CK19-NOUSE-DAG: [[CSVAL0]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}} // CK19-NOUSE-DAG: [[CSVAL0]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}}
// CK19-USE: call void [[CALL37:@.+]](i[[Z]] 11, i[[Z]] %{{[^,]+}}, [13 x double]* %{{[^,]+}}) // CK19-USE: call void [[CALL37:@.+]](i[[Z]] 11, i[[Z]] %{{[^,]+}}, [13 x double]* %{{[^,]+}})
// CK19-NOUSE: call void [[CALL37:@.+]]() // CK19-NOUSE: call void [[CALL37:@.+]]()
#pragma omp target map(mvlaas) #pragma omp target map(mvlaas)
{ {
#ifdef USE #ifdef USE
mvlaas[1][2][3]++; mvlaas[1][2][3]++;
#endif #endif
} }
// Region 38 // Region 38
// CK19-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[{{1|3}} x i{{.+}}]* [[MTYPE38]]{{.+}}, i8** null, i8** null) // CK19-USE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[SIZE38]]{{.+}}, {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE38]]{{.+}}, i8** null, i8** null)
// CK19-NOUSE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE38]]{{.+}}, i8** null, i8** null)
// CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK19-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// //
// CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]* // CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]*
// CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]* // CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CBP0]] // CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CBP0]]
// CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CP0]] // CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CP0]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S0]]
// //
// CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[Z]]* // CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[Z]]*
// CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[Z]]* // CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] [[VAR1:%.+]], i[[Z]]* [[CBP1]] // CK19-USE-DAG: store i[[Z]] [[VAR1:%.+]], i[[Z]]* [[CBP1]]
// CK19-USE-DAG: store i[[Z]] [[VAR11:%.+]], i[[Z]]* [[CP1]] // CK19-USE-DAG: store i[[Z]] [[VAR11:%.+]], i[[Z]]* [[CP1]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S1]]
// //
// CK19-USE-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK19-USE-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK19-USE-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [13 x double]** // CK19-USE-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [13 x double]**
// CK19-USE-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to [13 x double]** // CK19-USE-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to [13 x double]**
// CK19-USE-DAG: store [13 x double]* [[VAR2:%.+]], [13 x double]** [[CBP2]] // CK19-USE-DAG: store [13 x double]* [[VAR2:%.+]], [13 x double]** [[CBP2]]
// CK19-USE-DAG: store [13 x double]* [[SEC2:%.+]], [13 x double]** [[CP2]] // CK19-USE-DAG: store [13 x double]* [[SEC2:%.+]], [13 x double]** [[CP2]]
// CK19-USE-DAG: store i64 [[CSVAL2:%[^,]+]], i64* [[S2]] // CK19-USE-DAG: store i64 [[CSVAL2:%[^,]+]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE38]], i32 0, i32 2),
// CK19-USE-DAG: [[SEC2]] = getelementptr {{.+}}[13 x double]* [[VAR2]], i[[Z]] [[SEC22:%[^,]+]] // CK19-USE-DAG: [[SEC2]] = getelementptr {{.+}}[13 x double]* [[VAR2]], i[[Z]] [[SEC22:%[^,]+]]
// CK19-USE-DAG: [[SEC22]] = mul nsw i[[Z]] 0, %{{[^,]+}} // CK19-USE-DAG: [[SEC22]] = mul nsw i[[Z]] 0, %{{[^,]+}}
// CK19-USE-DAG: [[CSVAL2]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}} // CK19-USE-DAG: [[CSVAL2]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}}
// CK19-NOUSE-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [13 x double]** // CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [13 x double]**
// CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [13 x double]** // CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [13 x double]**
// CK19-NOUSE-DAG: store [13 x double]* [[VAR0:%.+]], [13 x double]** [[CBP0]] // CK19-NOUSE-DAG: store [13 x double]* [[VAR0:%.+]], [13 x double]** [[CBP0]]
// CK19-NOUSE-DAG: store [13 x double]* [[SEC0:%.+]], [13 x double]** [[CP0]] // CK19-NOUSE-DAG: store [13 x double]* [[SEC0:%.+]], [13 x double]** [[CP0]]
// CK19-NOUSE-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]] // CK19-NOUSE-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]]
// CK19-NOUSE-DAG: [[SEC0]] = getelementptr {{.+}}[13 x double]* [[VAR0]], i[[Z]] [[SEC00:%[^,]+]] // CK19-NOUSE-DAG: [[SEC0]] = getelementptr {{.+}}[13 x double]* [[VAR0]], i[[Z]] [[SEC00:%[^,]+]]
// CK19-NOUSE-DAG: [[SEC00]] = mul nsw i[[Z]] 0, %{{[^,]+}} // CK19-NOUSE-DAG: [[SEC00]] = mul nsw i[[Z]] 0, %{{[^,]+}}
// CK19-NOUSE-DAG: [[CSVAL0]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}} // CK19-NOUSE-DAG: [[CSVAL0]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}}
// CK19-USE: call void [[CALL38:@.+]](i[[Z]] 11, i[[Z]] %{{[^,]+}}, [13 x double]* %{{[^,]+}}) // CK19-USE: call void [[CALL38:@.+]](i[[Z]] 11, i[[Z]] %{{[^,]+}}, [13 x double]* %{{[^,]+}})
// CK19-NOUSE: call void [[CALL38:@.+]]() // CK19-NOUSE: call void [[CALL38:@.+]]()
#pragma omp target map(mvlaas[:]) #pragma omp target map(mvlaas[:])
{ {
#ifdef USE #ifdef USE
mvlaas[1][2][3]++; mvlaas[1][2][3]++;
#endif #endif
} }
// Region 39 // Region 39
// CK19-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[{{1|3}} x i{{.+}}]* [[MTYPE39]]{{.+}}, i8** null, i8** null) // CK19-USE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[SIZE39]]{{.+}}, {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE39]]{{.+}}, i8** null, i8** null)
// CK19-NOUSE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE39]]{{.+}}, i8** null, i8** null)
// CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK19-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// //
// CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]* // CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]*
// CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]* // CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CBP0]] // CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CBP0]]
// CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CP0]] // CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CP0]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S0]]
// //
// CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[Z]]* // CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[Z]]*
// CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[Z]]* // CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] [[VAR1:%.+]], i[[Z]]* [[CBP1]] // CK19-USE-DAG: store i[[Z]] [[VAR1:%.+]], i[[Z]]* [[CBP1]]
// CK19-USE-DAG: store i[[Z]] [[VAR11:%.+]], i[[Z]]* [[CP1]] // CK19-USE-DAG: store i[[Z]] [[VAR11:%.+]], i[[Z]]* [[CP1]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S1]]
// //
// CK19-USE-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK19-USE-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK19-USE-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [13 x double]** // CK19-USE-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [13 x double]**
// CK19-USE-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to [13 x double]** // CK19-USE-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to [13 x double]**
// CK19-USE-DAG: store [13 x double]* [[VAR2:%.+]], [13 x double]** [[CBP2]] // CK19-USE-DAG: store [13 x double]* [[VAR2:%.+]], [13 x double]** [[CBP2]]
// CK19-USE-DAG: store [13 x double]* [[SEC2:%.+]], [13 x double]** [[CP2]] // CK19-USE-DAG: store [13 x double]* [[SEC2:%.+]], [13 x double]** [[CP2]]
// CK19-USE-DAG: store i64 [[CSVAL2:%[^,]+]], i64* [[S2]] // CK19-USE-DAG: store i64 [[CSVAL2:%[^,]+]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE39]], i32 0, i32 2),
// CK19-USE-DAG: [[SEC2]] = getelementptr {{.+}}[13 x double]* [[VAR2]], i[[Z]] [[SEC22:%[^,]+]] // CK19-USE-DAG: [[SEC2]] = getelementptr {{.+}}[13 x double]* [[VAR2]], i[[Z]] [[SEC22:%[^,]+]]
// CK19-USE-DAG: [[SEC22]] = mul nsw i[[Z]] 0, %{{[^,]+}} // CK19-USE-DAG: [[SEC22]] = mul nsw i[[Z]] 0, %{{[^,]+}}
// CK19-USE-DAG: [[CSVAL2]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}} // CK19-USE-DAG: [[CSVAL2]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}}
// CK19-NOUSE-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [13 x double]** // CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [13 x double]**
// CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [13 x double]** // CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [13 x double]**
// CK19-NOUSE-DAG: store [13 x double]* [[VAR0:%.+]], [13 x double]** [[CBP0]] // CK19-NOUSE-DAG: store [13 x double]* [[VAR0:%.+]], [13 x double]** [[CBP0]]
// CK19-NOUSE-DAG: store [13 x double]* [[SEC0:%.+]], [13 x double]** [[CP0]] // CK19-NOUSE-DAG: store [13 x double]* [[SEC0:%.+]], [13 x double]** [[CP0]]
// CK19-NOUSE-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]] // CK19-NOUSE-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]]
// CK19-NOUSE-DAG: [[SEC0]] = getelementptr {{.+}}[13 x double]* [[VAR0]], i[[Z]] [[SEC00:%[^,]+]] // CK19-NOUSE-DAG: [[SEC0]] = getelementptr {{.+}}[13 x double]* [[VAR0]], i[[Z]] [[SEC00:%[^,]+]]
// CK19-NOUSE-DAG: [[SEC00]] = mul nsw i[[Z]] 0, %{{[^,]+}} // CK19-NOUSE-DAG: [[SEC00]] = mul nsw i[[Z]] 0, %{{[^,]+}}
// CK19-NOUSE-DAG: [[CSVAL0]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}} // CK19-NOUSE-DAG: [[CSVAL0]] = {{mul nuw i64 %[^,]+, 104|sext i32 .+ to i64}}
// CK19-USE: call void [[CALL39:@.+]](i[[Z]] 11, i[[Z]] %{{[^,]+}}, [13 x double]* %{{[^,]+}}) // CK19-USE: call void [[CALL39:@.+]](i[[Z]] 11, i[[Z]] %{{[^,]+}}, [13 x double]* %{{[^,]+}})
// CK19-NOUSE: call void [[CALL39:@.+]]() // CK19-NOUSE: call void [[CALL39:@.+]]()
#pragma omp target map(mvlaas[:][:][:]) #pragma omp target map(mvlaas[:][:][:])
{ {
#ifdef USE #ifdef USE
mvlaas[1][2][3]++; mvlaas[1][2][3]++;
#endif #endif
} }
// Region 40 // Region 40
// CK19-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[{{1|3}} x i{{.+}}]* [[MTYPE40]]{{.+}}, i8** null, i8** null) // CK19-USE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[SIZE40]]{{.+}}, {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE40]]{{.+}}, i8** null, i8** null)
// CK19-NOUSE-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 {{1|3}}, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE40]]{{.+}}, i8** null, i8** null)
// CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK19-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK19-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK19-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// //
// CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-USE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]* // CK19-USE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to i[[Z]]*
// CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]* // CK19-USE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CBP0]] // CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CBP0]]
// CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CP0]] // CK19-USE-DAG: store i[[Z]] 11, i[[Z]]* [[CP0]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S0]]
// //
// CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK19-USE-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[Z]]* // CK19-USE-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to i[[Z]]*
// CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[Z]]* // CK19-USE-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i[[Z]]*
// CK19-USE-DAG: store i[[Z]] [[VAR1:%.+]], i[[Z]]* [[CBP1]] // CK19-USE-DAG: store i[[Z]] [[VAR1:%.+]], i[[Z]]* [[CBP1]]
// CK19-USE-DAG: store i[[Z]] [[VAR11:%.+]], i[[Z]]* [[CP1]] // CK19-USE-DAG: store i[[Z]] [[VAR11:%.+]], i[[Z]]* [[CP1]]
// CK19-USE-DAG: store i64 {{8|4}}, i64* [[S1]]
// //
// CK19-USE-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK19-USE-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK19-USE-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK19-USE-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [13 x double]** // CK19-USE-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [13 x double]**
// CK19-USE-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to [13 x double]** // CK19-USE-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to [13 x double]**
// CK19-USE-DAG: store [13 x double]* [[VAR2:%.+]], [13 x double]** [[CBP2]] // CK19-USE-DAG: store [13 x double]* [[VAR2:%.+]], [13 x double]** [[CBP2]]
// CK19-USE-DAG: store [13 x double]* [[SEC2:%.+]], [13 x double]** [[CP2]] // CK19-USE-DAG: store [13 x double]* [[SEC2:%.+]], [13 x double]** [[CP2]]
// CK19-USE-DAG: store i64 [[CSVAL2:%[^,]+]], i64* [[S2]] // CK19-USE-DAG: store i64 [[CSVAL2:%[^,]+]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE40]], i32 0, i32 2),
// CK19-USE-DAG: [[SEC2]] = getelementptr {{.+}}[13 x double]* [[SEC22:%[^,]+]], i[[Z]] 0 // CK19-USE-DAG: [[SEC2]] = getelementptr {{.+}}[13 x double]* [[SEC22:%[^,]+]], i[[Z]] 0
// CK19-USE-DAG: [[SEC22]] = getelementptr {{.+}}[13 x double]* [[VAR2]], i[[Z]] [[SEC222:%[^,]+]] // CK19-USE-DAG: [[SEC22]] = getelementptr {{.+}}[13 x double]* [[VAR2]], i[[Z]] [[SEC222:%[^,]+]]
// CK19-USE-DAG: [[SEC222]] = mul nsw i[[Z]] 1, %{{[^,]+}} // CK19-USE-DAG: [[SEC222]] = mul nsw i[[Z]] 1, %{{[^,]+}}
// CK19-NOUSE-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0 // CK19-NOUSE-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [13 x double]** // CK19-NOUSE-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [13 x double]**
// CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [13 x double]** // CK19-NOUSE-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [13 x double]**
// CK19-NOUSE-DAG: store [13 x double]* [[VAR0:%.+]], [13 x double]** [[CBP0]] // CK19-NOUSE-DAG: store [13 x double]* [[VAR0:%.+]], [13 x double]** [[CBP0]]
// CK19-NOUSE-DAG: store [13 x double]* [[SEC0:%.+]], [13 x double]** [[CP0]] // CK19-NOUSE-DAG: store [13 x double]* [[SEC0:%.+]], [13 x double]** [[CP0]]
// CK19-NOUSE-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]] // CK19-NOUSE-DAG: store i64 [[CSVAL0:%[^,]+]], i64* [[S0]]
▲ Show 20 Lines • Show All 209 Lines • Show Last 20 Lines

clang/test/OpenMP/target_map_codegen_20.cpp

Show First 20 Lines • Show All 68 Lines • ▼ Show 20 Lines
// CK21-NOUSE: [[MTYPE00:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK21-NOUSE: [[MTYPE00:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK21-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK21-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK21: [[SIZE01:@.+]] = private {{.*}}constant [1 x i64] [i64 492] // CK21: [[SIZE01:@.+]] = private {{.*}}constant [1 x i64] [i64 492]
// CK21-USE: [[MTYPE01:@.+]] = private {{.*}}constant [1 x i64] [i64 35] // CK21-USE: [[MTYPE01:@.+]] = private {{.*}}constant [1 x i64] [i64 35]
// CK21-NOUSE: [[MTYPE01:@.+]] = private {{.*}}constant [1 x i64] [i64 3] // CK21-NOUSE: [[MTYPE01:@.+]] = private {{.*}}constant [1 x i64] [i64 3]
// CK21-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK21-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK21: [[SIZE02:@.+]] = private {{.*}}global [2 x i64] [i64 0, i64 500]
// CK21-USE: [[MTYPE02:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710674] // CK21-USE: [[MTYPE02:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710674]
// CK21-NOUSE: [[MTYPE02:@.+]] = private {{.*}}constant [2 x i64] [i64 0, i64 281474976710674] // CK21-NOUSE: [[MTYPE02:@.+]] = private {{.*}}constant [2 x i64] [i64 0, i64 281474976710674]
// CK21-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK21-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK21: [[SIZE03:@.+]] = private {{.*}}constant [1 x i64] [i64 492] // CK21: [[SIZE03:@.+]] = private {{.*}}constant [1 x i64] [i64 492]
// CK21-USE: [[MTYPE03:@.+]] = private {{.*}}constant [1 x i64] [i64 34] // CK21-USE: [[MTYPE03:@.+]] = private {{.*}}constant [1 x i64] [i64 34]
// CK21-NOUSE: [[MTYPE03:@.+]] = private {{.*}}constant [1 x i64] [i64 2] // CK21-NOUSE: [[MTYPE03:@.+]] = private {{.*}}constant [1 x i64] [i64 2]
// CK21-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK21-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK21: [[SIZE04:@.+]] = private {{.*}}constant [1 x i64] [i64 4] // CK21: [[SIZE04:@.+]] = private {{.*}}constant [1 x i64] [i64 4]
// CK21-USE: [[MTYPE04:@.+]] = private {{.*}}constant [1 x i64] [i64 34] // CK21-USE: [[MTYPE04:@.+]] = private {{.*}}constant [1 x i64] [i64 34]
// CK21-NOUSE: [[MTYPE04:@.+]] = private {{.*}}constant [1 x i64] [i64 2] // CK21-NOUSE: [[MTYPE04:@.+]] = private {{.*}}constant [1 x i64] [i64 2]
// CK21-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK21-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK21: [[SIZE05:@.+]] = private {{.*}}global [3 x i64] [i64 0, i64 4, i64 4]
// CK21-USE: [[MTYPE05:@.+]] = private {{.*}}constant [3 x i64] [i64 32, i64 281474976710659, i64 281474976710659] // CK21-USE: [[MTYPE05:@.+]] = private {{.*}}constant [3 x i64] [i64 32, i64 281474976710659, i64 281474976710659]
// CK21-NOUSE: [[MTYPE05:@.+]] = private {{.*}}constant [3 x i64] [i64 0, i64 281474976710659, i64 281474976710659] // CK21-NOUSE: [[MTYPE05:@.+]] = private {{.*}}constant [3 x i64] [i64 0, i64 281474976710659, i64 281474976710659]
// CK21-LABEL: explicit_maps_template_args_and_members{{.*}}( // CK21-LABEL: explicit_maps_template_args_and_members{{.*}}(
template <int X, typename T> template <int X, typename T>
struct CC { struct CC {
T A; T A;
▲ Show 20 Lines • Show All 46 Lines • ▼ Show 20 Lines#endif
#pragma omp target map(lb[:X]) #pragma omp target map(lb[:X])
{ {
#ifdef USE #ifdef USE
lb[4] += 1; lb[4] += 1;
#endif #endif
} }
// Region 02 // Region 02
// CK21-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE02]]{{.+}}, i8** null, i8** null) // CK21-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE02]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE02]]{{.+}}, i8** null, i8** null)
// CK21-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK21-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK21-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK21-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK21-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK21-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK21-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK21-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK21-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK21-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK21-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK21-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK21-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to float*** // CK21-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to float***
// CK21-DAG: store [[ST]]* [[VAR0:%.+]], [[ST]]** [[CBP0]] // CK21-DAG: store [[ST]]* [[VAR0:%.+]], [[ST]]** [[CBP0]]
// CK21-DAG: store float** [[SEC0:%.+]], float*** [[CP0]] // CK21-DAG: store float** [[SEC0:%.+]], float*** [[CP0]]
// CK21-DAG: store i64 {{%.+}}, i64* [[S0]] // CK21-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE02]], i32 0, i32 0),
// CK21-DAG: [[SEC0]] = getelementptr {{.*}}[[ST]]* [[VAR0]], i{{.+}} 0, i{{.+}} 2 // CK21-DAG: [[SEC0]] = getelementptr {{.*}}[[ST]]* [[VAR0]], i{{.+}} 0, i{{.+}} 2
// CK21-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK21-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK21-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK21-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK21-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK21-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to float*** // CK21-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to float***
// CK21-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to float** // CK21-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to float**
// CK21-DAG: store float** [[SEC0]], float*** [[CBP1]] // CK21-DAG: store float** [[SEC0]], float*** [[CBP1]]
// CK21-DAG: store float* [[SEC1:%.+]], float** [[CP1]] // CK21-DAG: store float* [[SEC1:%.+]], float** [[CP1]]
// CK21-DAG: store i64 {{.+}}, i64* [[S1]]
// CK21-DAG: [[SEC1]] = getelementptr {{.*}}float* [[RVAR1:%[^,]+]], i{{.+}} 123 // CK21-DAG: [[SEC1]] = getelementptr {{.*}}float* [[RVAR1:%[^,]+]], i{{.+}} 123
// CK21-DAG: [[RVAR1]] = load float*, float** [[SEC1_:%[^,]+]] // CK21-DAG: [[RVAR1]] = load float*, float** [[SEC1_:%[^,]+]]
// CK21-DAG: [[SEC1_]] = getelementptr {{.*}}[[ST]]* [[VAR0]], i{{.+}} 0, i{{.+}} 2 // CK21-DAG: [[SEC1_]] = getelementptr {{.*}}[[ST]]* [[VAR0]], i{{.+}} 0, i{{.+}} 2
// CK21-USE: call void [[CALL02:@.+]]([[ST]]* {{[^,]+}}) // CK21-USE: call void [[CALL02:@.+]]([[ST]]* {{[^,]+}})
// CK21-NOUSE: call void [[CALL02:@.+]]() // CK21-NOUSE: call void [[CALL02:@.+]]()
#pragma omp target map(from:B[X:X+2]) #pragma omp target map(from:B[X:X+2])
{ {
▲ Show 20 Lines • Show All 41 Lines • ▼ Show 20 Lines#endif
{ {
#ifdef USE #ifdef USE
arg +=1; arg +=1;
#endif #endif
} }
// Make sure the extra flag is passed to the second map. // Make sure the extra flag is passed to the second map.
// Region 05 // Region 05
// CK21-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE05]]{{.+}}, i8** null, i8** null) // CK21-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[SIZE05]]{{.+}}, {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE05]]{{.+}}, i8** null, i8** null)
// CK21-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK21-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK21-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK21-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK21-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK21-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK21-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK21-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK21-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK21-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK21-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK21-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK21-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK21-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK21-DAG: store [[ST]]* [[VAR0:%.+]], [[ST]]** [[CBP0]] // CK21-DAG: store [[ST]]* [[VAR0:%.+]], [[ST]]** [[CBP0]]
// CK21-DAG: store i32* [[SEC0:%.+]], i32** [[CP0]] // CK21-DAG: store i32* [[SEC0:%.+]], i32** [[CP0]]
// CK21-DAG: store i64 {{%.+}}, i64* [[S0]] // CK21-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE05]], i32 0, i32 0),
// CK21-DAG: [[SEC0]] = getelementptr {{.*}}[[ST]]* [[VAR0]], i{{.+}} 0, i{{.+}} 0 // CK21-DAG: [[SEC0]] = getelementptr {{.*}}[[ST]]* [[VAR0]], i{{.+}} 0, i{{.+}} 0
// CK21-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK21-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK21-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK21-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK21-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK21-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]** // CK21-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]**
// CK21-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK21-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK21-DAG: store [[ST]]* [[VAR0]], [[ST]]** [[CBP1]] // CK21-DAG: store [[ST]]* [[VAR0]], [[ST]]** [[CBP1]]
// CK21-DAG: store i32* [[SEC0]], i32** [[CP1]] // CK21-DAG: store i32* [[SEC0]], i32** [[CP1]]
// CK21-DAG: store i64 {{.+}}, i64* [[S1]]
// CK21-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK21-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK21-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK21-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK21-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK21-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]** // CK21-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]**
// CK21-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32** // CK21-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32**
// CK21-DAG: store [[ST]]* [[VAR2:%.+]], [[ST]]** [[CBP2]] // CK21-DAG: store [[ST]]* [[VAR2:%.+]], [[ST]]** [[CBP2]]
// CK21-DAG: store i32* [[SEC2:%.+]], i32** [[CP2]] // CK21-DAG: store i32* [[SEC2:%.+]], i32** [[CP2]]
// CK21-DAG: store i64 {{.+}}, i64* [[S2]]
// CK21-DAG: [[SEC2]] = getelementptr {{.*}}[[ST]]* [[VAR2]], i{{.+}} 0, i{{.+}} 1 // CK21-DAG: [[SEC2]] = getelementptr {{.*}}[[ST]]* [[VAR2]], i{{.+}} 0, i{{.+}} 1
// CK21-USE: call void [[CALL05:@.+]]([[ST]]* {{[^,]+}}) // CK21-USE: call void [[CALL05:@.+]]([[ST]]* {{[^,]+}})
// CK21-NOUSE: call void [[CALL05:@.+]]() // CK21-NOUSE: call void [[CALL05:@.+]]()
#pragma omp target map(A, A2) #pragma omp target map(A, A2)
{ {
#ifdef USE #ifdef USE
A += 1; A += 1;
Show All 20 Lines

clang/test/OpenMP/target_map_codegen_23.cpp

Show First 20 Lines • Show All 67 Lines • ▼ Show 20 Lines
// CK24: [[SIZE14:@.+]] = private {{.*}}constant [1 x i64] [i64 {{48|56}}] // CK24: [[SIZE14:@.+]] = private {{.*}}constant [1 x i64] [i64 {{48|56}}]
// CK24: [[MTYPE14:@.+]] = private {{.*}}constant [1 x i64] [i64 35] // CK24: [[MTYPE14:@.+]] = private {{.*}}constant [1 x i64] [i64 35]
// CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK24: [[SIZE15:@.+]] = private {{.*}}constant [1 x i64] [i64 4] // CK24: [[SIZE15:@.+]] = private {{.*}}constant [1 x i64] [i64 4]
// CK24: [[MTYPE15:@.+]] = private {{.*}}constant [1 x i64] [i64 35] // CK24: [[MTYPE15:@.+]] = private {{.*}}constant [1 x i64] [i64 35]
// CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK24: [[SIZE16:@.+]] = private {{.*}}global [2 x i64] [i64 0, i64 20]
// CK24: [[MTYPE16:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710659] // CK24: [[MTYPE16:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710659]
// CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK24: [[SIZE17:@.+]] = private {{.*}}global [2 x i64] [i64 0, i64 {{3560|2880}}]
// CK24: [[MTYPE17:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710675] // CK24: [[MTYPE17:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710675]
// CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK24: [[SIZE18:@.+]] = private {{.*}}constant [1 x i64] [i64 4] // CK24: [[SIZE18:@.+]] = private {{.*}}constant [1 x i64] [i64 4]
// CK24: [[MTYPE18:@.+]] = private {{.*}}constant [1 x i64] [i64 35] // CK24: [[MTYPE18:@.+]] = private {{.*}}constant [1 x i64] [i64 35]
// CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK24: [[SIZE19:@.+]] = private {{.*}}global [3 x i64] [i64 0, i64 {{8|4}}, i64 4]
// CK24: [[MTYPE19:@.+]] = private {{.*}}constant [3 x i64] [i64 32, i64 281474976710659, i64 281474976710675] // CK24: [[MTYPE19:@.+]] = private {{.*}}constant [3 x i64] [i64 32, i64 281474976710659, i64 281474976710675]
// CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK24: [[SIZE20:@.+]] = private {{.*}}global [2 x i64] [i64 0, i64 4]
// CK24: [[MTYPE20:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710675] // CK24: [[MTYPE20:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710675]
// CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK24: [[SIZE21:@.+]] = private {{.*}}global [3 x i64] [i64 0, i64 {{8|4}}, i64 4]
// CK24: [[MTYPE21:@.+]] = private {{.*}}constant [3 x i64] [i64 32, i64 281474976710659, i64 281474976710675] // CK24: [[MTYPE21:@.+]] = private {{.*}}constant [3 x i64] [i64 32, i64 281474976710659, i64 281474976710675]
// CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK24: [[SIZE22:@.+]] = private {{.*}}global [2 x i64] [i64 0, i64 8]
// CK24: [[MTYPE22:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710659] // CK24: [[MTYPE22:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710659]
// CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK24: [[SIZE23:@.+]] = private {{.*}}global [3 x i64] [i64 0, i64 {{8|4}}, i64 8]
// CK24: [[MTYPE23:@.+]] = private {{.*}}constant [3 x i64] [i64 32, i64 281474976710659, i64 281474976710675] // CK24: [[MTYPE23:@.+]] = private {{.*}}constant [3 x i64] [i64 32, i64 281474976710659, i64 281474976710675]
// CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK24-LABEL: @.__omp_offloading_{{.*}}explicit_maps_struct_fields{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK24: [[SIZE24:@.+]] = private {{.*}}global [4 x i64] [i64 0, i64 {{8|4}}, i64 {{8|4}}, i64 4]
// CK24: [[MTYPE24:@.+]] = private {{.*}}constant [4 x i64] [i64 32, i64 281474976710672, i64 16, i64 19] // CK24: [[MTYPE24:@.+]] = private {{.*}}constant [4 x i64] [i64 32, i64 281474976710672, i64 16, i64 19]
// CK24-LABEL: explicit_maps_struct_fields // CK24-LABEL: explicit_maps_struct_fields
int explicit_maps_struct_fields(int a){ int explicit_maps_struct_fields(int a){
SC s; SC s;
SC *p; SC *p;
// Region 01 // Region 01
▲ Show 20 Lines • Show All 75 Lines • ▼ Show 20 Lines
// CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24: call void [[CALL15:@.+]]([[SC]]* {{[^,]+}}) // CK24: call void [[CALL15:@.+]]([[SC]]* {{[^,]+}})
#pragma omp target map(p->s.s.a) #pragma omp target map(p->s.s.a)
{ p->a++; } { p->a++; }
// Region 16 // Region 16
// CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE16]]{{.+}}, i8** null) // CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE16]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE16]]{{.+}}, i8** null)
// CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK24-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]** // CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]**
// CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]] // CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]]
// CK24-DAG: store i32* [[SEC0:%.+]], i32** [[CP0]] // CK24-DAG: store i32* [[SEC0:%.+]], i32** [[CP0]]
// CK24-DAG: store i64 {{%.+}}, i64* [[S0]] // CK24-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE16]], i32 0, i32 0),
// CK24-DAG: [[SEC0]] = getelementptr {{.*}}[10 x i32]* [[SEC00:%[^,]+]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[SEC0]] = getelementptr {{.*}}[10 x i32]* [[SEC00:%[^,]+]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[SEC00]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 3 // CK24-DAG: [[SEC00]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 3
// CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SC]]** // CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SC]]**
// CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK24-DAG: store [[SC]]* [[VAR0]], [[SC]]** [[CBP1]] // CK24-DAG: store [[SC]]* [[VAR0]], [[SC]]** [[CBP1]]
// CK24-DAG: store i32* [[SEC0]], i32** [[CP1]] // CK24-DAG: store i32* [[SEC0]], i32** [[CP1]]
// CK24-DAG: store i64 {{.+}}, i64* [[S1]]
// CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24: call void [[CALL16:@.+]]([[SC]]* {{[^,]+}}) // CK24: call void [[CALL16:@.+]]([[SC]]* {{[^,]+}})
#pragma omp target map(p->b[:5]) #pragma omp target map(p->b[:5])
{ p->a++; } { p->a++; }
// Region 17 // Region 17
// CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE17]]{{.+}}, i8** null) // CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE17]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE17]]{{.+}}, i8** null)
// CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK24-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]** // CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]**
// CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SB]]*** // CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SB]]***
// CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]] // CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]]
// CK24-DAG: store [[SB]]** [[SEC0:%.+]], [[SB]]*** [[CP0]] // CK24-DAG: store [[SB]]** [[SEC0:%.+]], [[SB]]*** [[CP0]]
// CK24-DAG: store i64 {{%.+}}, i64* [[S0]] // CK24-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE17]], i32 0, i32 0),
// CK24-DAG: [[SEC0]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[SEC0]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SB]]*** // CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SB]]***
// CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to [[SB]]** // CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to [[SB]]**
// CK24-DAG: store [[SB]]** [[SEC0]], [[SB]]*** [[CBP1]] // CK24-DAG: store [[SB]]** [[SEC0]], [[SB]]*** [[CBP1]]
// CK24-DAG: store [[SB]]* [[SEC1:%.+]], [[SB]]** [[CP1]] // CK24-DAG: store [[SB]]* [[SEC1:%.+]], [[SB]]** [[CP1]]
// CK24-DAG: store i64 {{.+}}, i64* [[S1]]
// CK24-DAG: [[SEC1]] = getelementptr {{.*}}[[SB]]* [[SEC11:%[^,]+]], i{{.+}} 0 // CK24-DAG: [[SEC1]] = getelementptr {{.*}}[[SB]]* [[SEC11:%[^,]+]], i{{.+}} 0
// CK24-DAG: [[SEC11]] = load [[SB]]*, [[SB]]** [[SEC111:%[^,]+]], // CK24-DAG: [[SEC11]] = load [[SB]]*, [[SB]]** [[SEC111:%[^,]+]],
// CK24-DAG: [[SEC111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[SEC111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR000]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR000]] = load [[SC]]*, [[SC]]** %{{.+}}
Show All 20 Lines
// CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24: call void [[CALL18:@.+]]([[SC]]* {{[^,]+}}) // CK24: call void [[CALL18:@.+]]([[SC]]* {{[^,]+}})
#pragma omp target map(p->s.sa[3].a) #pragma omp target map(p->s.sa[3].a)
{ p->a++; } { p->a++; }
// Region 19 // Region 19
// CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE19]]{{.+}}, i8** null) // CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[SIZE19]]{{.+}}, {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE19]]{{.+}}, i8** null)
// CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK24-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]** // CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]**
// CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SA]]*** // CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SA]]***
// CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]] // CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]]
// CK24-DAG: store [[SA]]** [[SEC0:%.+]], [[SA]]*** [[CP0]] // CK24-DAG: store [[SA]]** [[SEC0:%.+]], [[SA]]*** [[CP0]]
// CK24-DAG: store i64 {{%.+}}, i64* [[S0]] // CK24-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE19]], i32 0, i32 0),
// CK24-DAG: [[SEC0]] = getelementptr {{.*}}[10 x [[SA]]*]* [[SEC00:%[^,]+]], i{{.+}} 0, i{{.+}} 3 // CK24-DAG: [[SEC0]] = getelementptr {{.*}}[10 x [[SA]]*]* [[SEC00:%[^,]+]], i{{.+}} 0, i{{.+}} 3
// CK24-DAG: [[SEC00]] = getelementptr {{.*}}[[SB]]* [[SEC000:%[^,]+]], i{{.+}} 0, i{{.+}} 3 // CK24-DAG: [[SEC00]] = getelementptr {{.*}}[[SB]]* [[SEC000:%[^,]+]], i{{.+}} 0, i{{.+}} 3
// CK24-DAG: [[SEC000]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[SEC000]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SC]]** // CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SC]]**
// CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to [[SA]]*** // CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to [[SA]]***
// CK24-DAG: store [[SC]]* [[VAR0]], [[SC]]** [[CBP1]] // CK24-DAG: store [[SC]]* [[VAR0]], [[SC]]** [[CBP1]]
// CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CP1]] // CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CP1]]
// CK24-DAG: store i64 {{.+}}, i64* [[S1]]
// CK24-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[SA]]*** // CK24-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[SA]]***
// CK24-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32** // CK24-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32**
// CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CBP2]] // CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CBP2]]
// CK24-DAG: store i32* [[SEC1:%.+]], i32** [[CP2]] // CK24-DAG: store i32* [[SEC1:%.+]], i32** [[CP2]]
// CK24-DAG: [[SEC1]] = getelementptr {{.*}}[[SA]]* [[SEC11:%[^,]+]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[SEC1]] = getelementptr {{.*}}[[SA]]* [[SEC11:%[^,]+]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[SEC11]] = load [[SA]]*, [[SA]]** [[SEC111:%[^,]+]], // CK24-DAG: [[SEC11]] = load [[SA]]*, [[SA]]** [[SEC111:%[^,]+]],
// CK24-DAG: [[SEC111]] = getelementptr {{.*}}[10 x [[SA]]*]* [[SEC1111:%[^,]+]], i{{.+}} 0, i{{.+}} 3 // CK24-DAG: [[SEC111]] = getelementptr {{.*}}[10 x [[SA]]*]* [[SEC1111:%[^,]+]], i{{.+}} 0, i{{.+}} 3
// CK24-DAG: [[SEC1111]] = getelementptr {{.*}}[[SB]]* [[SEC11111:%[^,]+]], i{{.+}} 0, i{{.+}} 3 // CK24-DAG: [[SEC1111]] = getelementptr {{.*}}[[SB]]* [[SEC11111:%[^,]+]], i{{.+}} 0, i{{.+}} 3
// CK24-DAG: [[SEC11111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[SEC11111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR000]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR000]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24: call void [[CALL19:@.+]]([[SC]]* {{[^,]+}}) // CK24: call void [[CALL19:@.+]]([[SC]]* {{[^,]+}})
#pragma omp target map(p->s.sp[3]->a) #pragma omp target map(p->s.sp[3]->a)
{ p->a++; } { p->a++; }
// Region 20 // Region 20
// CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE20]]{{.+}}, i8** null) // CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE20]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE20]]{{.+}}, i8** null)
// CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK24-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]** // CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]**
// CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SB]]*** // CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SB]]***
// CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]] // CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]]
// CK24-DAG: store [[SB]]** [[SEC0:%.+]], [[SB]]*** [[CP0]] // CK24-DAG: store [[SB]]** [[SEC0:%.+]], [[SB]]*** [[CP0]]
// CK24-DAG: store i64 {{%.+}}, i64* [[S0]] // CK24-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE20]], i32 0, i32 0),
// CK24-DAG: [[SEC0]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[SEC0]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SB]]*** // CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SB]]***
// CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK24-DAG: store [[SB]]** [[SEC0]], [[SB]]*** [[CBP1]] // CK24-DAG: store [[SB]]** [[SEC0]], [[SB]]*** [[CBP1]]
// CK24-DAG: store i32* [[SEC1:%.+]], i32** [[CP1]] // CK24-DAG: store i32* [[SEC1:%.+]], i32** [[CP1]]
// CK24-DAG: [[SEC1]] = getelementptr {{.*}}[[SB]]* [[SEC11:%[^,]+]], i{{.+}} 0 // CK24-DAG: [[SEC1]] = getelementptr {{.*}}[[SB]]* [[SEC11:%[^,]+]], i{{.+}} 0
// CK24-DAG: [[SEC11]] = load [[SB]]*, [[SB]]** [[SEC111:%[^,]+]], // CK24-DAG: [[SEC11]] = load [[SB]]*, [[SB]]** [[SEC111:%[^,]+]],
// CK24-DAG: [[SEC111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[SEC111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR000]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR000]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24: call void [[CALL20:@.+]]([[SC]]* {{[^,]+}}) // CK24: call void [[CALL20:@.+]]([[SC]]* {{[^,]+}})
#pragma omp target map(p->p->a) #pragma omp target map(p->p->a)
{ p->a++; } { p->a++; }
// Region 21 // Region 21
// CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE21]]{{.+}}, i8** null) // CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[SIZE21]]{{.+}}, {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE21]]{{.+}}, i8** null)
// CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK24-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]** // CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]**
// CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SA]]*** // CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SA]]***
// CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]] // CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]]
// CK24-DAG: store [[SA]]** [[SEC0:%.+]], [[SA]]*** [[CP0]] // CK24-DAG: store [[SA]]** [[SEC0:%.+]], [[SA]]*** [[CP0]]
// CK24-DAG: store i64 {{%.+}}, i64* [[S0]] // CK24-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE21]], i32 0, i32 0),
// CK24-DAG: [[SEC0]] = getelementptr {{.*}}[[SB]]* [[SEC00:[^,]+]], i{{.+}} 0, i{{.+}} 4 // CK24-DAG: [[SEC0]] = getelementptr {{.*}}[[SB]]* [[SEC00:[^,]+]], i{{.+}} 0, i{{.+}} 4
// CK24-DAG: [[SEC00]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[SEC00]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SC]]** // CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SC]]**
// CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to [[SA]]*** // CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to [[SA]]***
// CK24-DAG: store [[SC]]* [[VAR0]], [[SC]]** [[CBP1]] // CK24-DAG: store [[SC]]* [[VAR0]], [[SC]]** [[CBP1]]
// CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CP1]] // CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CP1]]
// CK24-DAG: store i64 {{.+}}, i64* [[S1]]
// CK24-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[SA]]*** // CK24-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[SA]]***
// CK24-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32** // CK24-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32**
// CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CBP2]] // CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CBP2]]
// CK24-DAG: store i32* [[SEC1:%.+]], i32** [[CP2]] // CK24-DAG: store i32* [[SEC1:%.+]], i32** [[CP2]]
// CK24-DAG: store i64 {{.+}}, i64* [[S2]]
// CK24-DAG: [[SEC1]] = getelementptr {{.*}}[[SA]]* [[SEC11:%[^,]+]], i{{.+}} 0 // CK24-DAG: [[SEC1]] = getelementptr {{.*}}[[SA]]* [[SEC11:%[^,]+]], i{{.+}} 0
// CK24-DAG: [[SEC11]] = load [[SA]]*, [[SA]]** [[SEC111:%[^,]+]], // CK24-DAG: [[SEC11]] = load [[SA]]*, [[SA]]** [[SEC111:%[^,]+]],
// CK24-DAG: [[SEC111]] = getelementptr {{.*}}[[SB]]* [[SEC1111:[^,]+]], i{{.+}} 0, i{{.+}} 4 // CK24-DAG: [[SEC111]] = getelementptr {{.*}}[[SB]]* [[SEC1111:[^,]+]], i{{.+}} 0, i{{.+}} 4
// CK24-DAG: [[SEC1111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[SEC1111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR000]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR000]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24: call void [[CALL21:@.+]]([[SC]]* {{[^,]+}}) // CK24: call void [[CALL21:@.+]]([[SC]]* {{[^,]+}})
#pragma omp target map(p->s.p->a) #pragma omp target map(p->s.p->a)
{ p->a++; } { p->a++; }
// Region 22 // Region 22
// CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE22]]{{.+}}, i8** null) // CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE22]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE22]]{{.+}}, i8** null)
// CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK24-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]** // CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]**
// CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]] // CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]]
// CK24-DAG: store i32* [[SEC0:%.+]], i32** [[CP0]] // CK24-DAG: store i32* [[SEC0:%.+]], i32** [[CP0]]
// CK24-DAG: store i64 {{%.+}}, i64* [[S0]] // CK24-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE22]], i32 0, i32 0),
// CK24-DAG: [[SEC0]] = getelementptr {{.*}}[10 x i32]* [[SEC00:%[^,]+]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[SEC0]] = getelementptr {{.*}}[10 x i32]* [[SEC00:%[^,]+]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[SEC00]] = getelementptr {{.*}}[[SA]]* [[SEC000:%[^,]+]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[SEC00]] = getelementptr {{.*}}[[SA]]* [[SEC000:%[^,]+]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[SEC000]] = getelementptr {{.*}}[[SB]]* [[SEC0000:%[^,]+]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[SEC000]] = getelementptr {{.*}}[[SB]]* [[SEC0000:%[^,]+]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[SEC0000]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[SEC0000]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SC]]** // CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SC]]**
// CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK24-DAG: store [[SC]]* [[VAR0]], [[SC]]** [[CBP1]] // CK24-DAG: store [[SC]]* [[VAR0]], [[SC]]** [[CBP1]]
// CK24-DAG: store i32* [[SEC0]], i32** [[CP1]] // CK24-DAG: store i32* [[SEC0]], i32** [[CP1]]
// CK24-DAG: store i64 {{.+}}, i64* [[S1]]
// CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24: call void [[CALL22:@.+]]([[SC]]* {{[^,]+}}) // CK24: call void [[CALL22:@.+]]([[SC]]* {{[^,]+}})
#pragma omp target map(p->s.s.b[:2]) #pragma omp target map(p->s.s.b[:2])
{ p->a++; } { p->a++; }
// Region 23 // Region 23
// CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE23]]{{.+}}, i8** null) // CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[SIZE23]]{{.+}}, {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE23]]{{.+}}, i8** null)
// CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK24-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]** // CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]**
// CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SA]]*** // CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SA]]***
// CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]] // CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]]
// CK24-DAG: store [[SA]]** [[SEC0:%.+]], [[SA]]*** [[CP0]] // CK24-DAG: store [[SA]]** [[SEC0:%.+]], [[SA]]*** [[CP0]]
// CK24-DAG: store i64 {{%.+}}, i64* [[S0]] // CK24-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE23]], i32 0, i32 0),
// CK24-DAG: [[SEC0]] = getelementptr {{.*}}[[SB]]* [[SEC00:%[^,]+]], i{{.+}} 0, i{{.+}} 4 // CK24-DAG: [[SEC0]] = getelementptr {{.*}}[[SB]]* [[SEC00:%[^,]+]], i{{.+}} 0, i{{.+}} 4
// CK24-DAG: [[SEC00]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[SEC00]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SC]]** // CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SC]]**
// CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to [[SA]]*** // CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to [[SA]]***
// CK24-DAG: store [[SC]]* [[VAR0]], [[SC]]** [[CBP1]] // CK24-DAG: store [[SC]]* [[VAR0]], [[SC]]** [[CBP1]]
// CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CP1]] // CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CP1]]
// CK24-DAG: store i64 {{.+}}, i64* [[S1]]
// CK24-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[SA]]*** // CK24-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[SA]]***
// CK24-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32** // CK24-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32**
// CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CBP2]] // CK24-DAG: store [[SA]]** [[SEC0]], [[SA]]*** [[CBP2]]
// CK24-DAG: store i32* [[SEC1:%.+]], i32** [[CP2]] // CK24-DAG: store i32* [[SEC1:%.+]], i32** [[CP2]]
// CK24-DAG: store i64 {{.+}}, i64* [[S2]]
// CK24-DAG: [[SEC1]] = getelementptr {{.*}}[10 x i32]* [[SEC11:%[^,]+]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[SEC1]] = getelementptr {{.*}}[10 x i32]* [[SEC11:%[^,]+]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[SEC11]] = getelementptr {{.*}}[[SA]]* [[SEC111:%[^,]+]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[SEC11]] = getelementptr {{.*}}[[SA]]* [[SEC111:%[^,]+]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[SEC111]] = load [[SA]]*, [[SA]]** [[SEC1111:%[^,]+]], // CK24-DAG: [[SEC111]] = load [[SA]]*, [[SA]]** [[SEC1111:%[^,]+]],
// CK24-DAG: [[SEC1111]] = getelementptr {{.*}}[[SB]]* [[SEC11111:%[^,]+]], i{{.+}} 0, i{{.+}} 4 // CK24-DAG: [[SEC1111]] = getelementptr {{.*}}[[SB]]* [[SEC11111:%[^,]+]], i{{.+}} 0, i{{.+}} 4
// CK24-DAG: [[SEC11111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[SEC11111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR0]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR00]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24-DAG: [[VAR000]] = load [[SC]]*, [[SC]]** %{{.+}} // CK24-DAG: [[VAR000]] = load [[SC]]*, [[SC]]** %{{.+}}
// CK24: call void [[CALL23:@.+]]([[SC]]* {{[^,]+}}) // CK24: call void [[CALL23:@.+]]([[SC]]* {{[^,]+}})
#pragma omp target map(p->s.p->b[:2]) #pragma omp target map(p->s.p->b[:2])
{ p->a++; } { p->a++; }
// Region 24 // Region 24
// CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 4, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[MTYPE24]]{{.+}}, i8** null) // CK24-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 4, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[SIZE24]]{{.+}}, {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[MTYPE24]]{{.+}}, i8** null)
// CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK24-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK24-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK24-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]** // CK24-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SC]]**
// CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SB]]*** // CK24-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SB]]***
// CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]] // CK24-DAG: store [[SC]]* [[VAR0:%.+]], [[SC]]** [[CBP0]]
// CK24-DAG: store [[SB]]** [[SEC0:%.+]], [[SB]]*** [[CP0]] // CK24-DAG: store [[SB]]** [[SEC0:%.+]], [[SB]]*** [[CP0]]
// CK24-DAG: store i64 {{%.+}}, i64* [[S0]] // CK24-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE24]], i32 0, i32 0),
// CK24-DAG: [[SEC0]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[SEC0]] = getelementptr {{.*}}[[SC]]* [[VAR00:%.+]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SB]]*** // CK24-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SB]]***
// CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to [[SA]]*** // CK24-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to [[SA]]***
// CK24-DAG: store [[SB]]** [[SEC0]], [[SB]]*** [[CBP1]] // CK24-DAG: store [[SB]]** [[SEC0]], [[SB]]*** [[CBP1]]
// CK24-DAG: store [[SA]]** [[SEC1:%.+]], [[SA]]*** [[CP1]] // CK24-DAG: store [[SA]]** [[SEC1:%.+]], [[SA]]*** [[CP1]]
// CK24-DAG: store i64 {{.+}}, i64* [[S1]]
// CK24-DAG: [[SEC1]] = getelementptr {{.*}}[[SB]]* [[SEC11:%[^,]+]], i{{.+}} 0, i{{.+}} 4 // CK24-DAG: [[SEC1]] = getelementptr {{.*}}[[SB]]* [[SEC11:%[^,]+]], i{{.+}} 0, i{{.+}} 4
// CK24-DAG: [[SEC11]] = load [[SB]]*, [[SB]]** [[SEC111:%[^,]+]], // CK24-DAG: [[SEC11]] = load [[SB]]*, [[SB]]** [[SEC111:%[^,]+]],
// CK24-DAG: [[SEC111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[SEC111]] = getelementptr {{.*}}[[SC]]* [[VAR000:%.+]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[SA]]*** // CK24-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[SA]]***
// CK24-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to [[SA]]*** // CK24-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to [[SA]]***
// CK24-DAG: store [[SA]]** [[SEC1]], [[SA]]*** [[CBP2]] // CK24-DAG: store [[SA]]** [[SEC1]], [[SA]]*** [[CBP2]]
// CK24-DAG: store [[SA]]** [[SEC2:%.+]], [[SA]]*** [[CP2]] // CK24-DAG: store [[SA]]** [[SEC2:%.+]], [[SA]]*** [[CP2]]
// CK24-DAG: store i64 {{.+}}, i64* [[S2]]
// CK24-DAG: [[SEC2]] = getelementptr {{.*}}[[SA]]* [[SEC22:%[^,]+]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[SEC2]] = getelementptr {{.*}}[[SA]]* [[SEC22:%[^,]+]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[SEC22]] = load [[SA]]*, [[SA]]** [[SEC222:%[^,]+]], // CK24-DAG: [[SEC22]] = load [[SA]]*, [[SA]]** [[SEC222:%[^,]+]],
// CK24-DAG: [[SEC222]] = getelementptr {{.*}}[[SB]]* [[SEC2222:%[^,]+]], i{{.+}} 0, i{{.+}} 4 // CK24-DAG: [[SEC222]] = getelementptr {{.*}}[[SB]]* [[SEC2222:%[^,]+]], i{{.+}} 0, i{{.+}} 4
// CK24-DAG: [[SEC2222]] = load [[SB]]*, [[SB]]** [[SEC22222:%[^,]+]], // CK24-DAG: [[SEC2222]] = load [[SB]]*, [[SB]]** [[SEC22222:%[^,]+]],
// CK24-DAG: [[SEC22222]] = getelementptr {{.*}}[[SC]]* [[VAR0000:%.+]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[SEC22222]] = getelementptr {{.*}}[[SC]]* [[VAR0000:%.+]], i{{.+}} 0, i{{.+}} 2
// CK24-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3 // CK24-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3
// CK24-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3 // CK24-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3
// CK24-DAG: [[S3:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 3
// CK24-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to [[SA]]*** // CK24-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to [[SA]]***
// CK24-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to i32** // CK24-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to i32**
// CK24-DAG: store [[SA]]** [[SEC2]], [[SA]]*** [[CBP3]] // CK24-DAG: store [[SA]]** [[SEC2]], [[SA]]*** [[CBP3]]
// CK24-DAG: store i32* [[SEC3:%.+]], i32** [[CP3]] // CK24-DAG: store i32* [[SEC3:%.+]], i32** [[CP3]]
// CK24-DAG: store i64 {{.+}}, i64* [[S3]]
// CK24-DAG: [[SEC3]] = getelementptr {{.*}}[[SA]]* [[SEC33:%[^,]+]], i{{.+}} 0, i{{.+}} 0 // CK24-DAG: [[SEC3]] = getelementptr {{.*}}[[SA]]* [[SEC33:%[^,]+]], i{{.+}} 0, i{{.+}} 0
// CK24-DAG: [[SEC33]] = load [[SA]]*, [[SA]]** [[SEC333:%[^,]+]], // CK24-DAG: [[SEC33]] = load [[SA]]*, [[SA]]** [[SEC333:%[^,]+]],
// CK24-DAG: [[SEC333]] = getelementptr {{.*}}[[SA]]* [[SEC3333:%[^,]+]], i{{.+}} 0, i{{.+}} 1 // CK24-DAG: [[SEC333]] = getelementptr {{.*}}[[SA]]* [[SEC3333:%[^,]+]], i{{.+}} 0, i{{.+}} 1
// CK24-DAG: [[SEC3333]] = load [[SA]]*, [[SA]]** [[SEC33333:%[^,]+]], // CK24-DAG: [[SEC3333]] = load [[SA]]*, [[SA]]** [[SEC33333:%[^,]+]],
// CK24-DAG: [[SEC33333]] = getelementptr {{.*}}[[SB]]* [[SEC333333:%[^,]+]], i{{.+}} 0, i{{.+}} 4 // CK24-DAG: [[SEC33333]] = getelementptr {{.*}}[[SB]]* [[SEC333333:%[^,]+]], i{{.+}} 0, i{{.+}} 4
// CK24-DAG: [[SEC333333]] = load [[SB]]*, [[SB]]** [[SEC3333333:%[^,]+]], // CK24-DAG: [[SEC333333]] = load [[SB]]*, [[SB]]** [[SEC3333333:%[^,]+]],
// CK24-DAG: [[SEC3333333]] = getelementptr {{.*}}[[SC]]* [[VAR00000:%.+]], i{{.+}} 0, i{{.+}} 2 // CK24-DAG: [[SEC3333333]] = getelementptr {{.*}}[[SC]]* [[VAR00000:%.+]], i{{.+}} 0, i{{.+}} 2
Show All 28 Lines

clang/test/OpenMP/target_map_codegen_28.cpp

Show All 31 Lines
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY28 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY28 %s
// SIMD-ONLY28-NOT: {{__kmpc|__tgt}} // SIMD-ONLY28-NOT: {{__kmpc|__tgt}}
#ifdef CK29 #ifdef CK29
// CK29: [[SSA:%.+]] = type { double*, double** } // CK29: [[SSA:%.+]] = type { double*, double** }
// CK29: [[SSB:%.+]] = type { [[SSA]]*, [[SSA]]** } // CK29: [[SSB:%.+]] = type { [[SSA]]*, [[SSA]]** }
// CK29-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK29-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK29: [[SIZE00:@.+]] = private {{.*}}global [2 x i64] [i64 0, i64 80]
// CK29: [[MTYPE00:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710675] // CK29: [[MTYPE00:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710675]
// CK29-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK29-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK29: [[SIZE01:@.+]] = private {{.*}}global [3 x i64] [i64 0, i64 {{8|4}}, i64 80]
// CK29: [[MTYPE01:@.+]] = private {{.*}}constant [3 x i64] [i64 32, i64 281474976710672, i64 19] // CK29: [[MTYPE01:@.+]] = private {{.*}}constant [3 x i64] [i64 32, i64 281474976710672, i64 19]
// CK29-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK29-LABEL: @.__omp_offloading_{{.*}}foo{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK29: [[SIZE02:@.+]] = private {{.*}}global [2 x i64] [i64 0, i64 80]
// CK29: [[MTYPE02:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710675] // CK29: [[MTYPE02:@.+]] = private {{.*}}constant [2 x i64] [i64 32, i64 281474976710675]
struct SSA{ struct SSA{
double *p; double *p;
double *&pr; double *&pr;
SSA(double *&pr) : pr(pr) {} SSA(double *&pr) : pr(pr) {}
}; };
struct SSB{ struct SSB{
SSA *p; SSA *p;
SSA *&pr; SSA *&pr;
SSB(SSA *&pr) : pr(pr) {} SSB(SSA *&pr) : pr(pr) {}
// CK29-LABEL: define {{.+}}foo // CK29-LABEL: define {{.+}}foo
void foo() { void foo() {
// Region 00 // Region 00
// CK29-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[Z:64|32]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null, i8** null) // CK29-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null, i8** null)
// CK29-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK29-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK29-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK29-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK29-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK29-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK29-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK29-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK29-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK29-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK29-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SSB]]** // CK29-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SSB]]**
// CK29-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SSA]]*** // CK29-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SSA]]***
// CK29-DAG: store [[SSB]]* [[VAR0:%.+]], [[SSB]]** [[CBP0]] // CK29-DAG: store [[SSB]]* [[VAR0:%.+]], [[SSB]]** [[CBP0]]
// CK29-DAG: store [[SSA]]** [[VAR00:%.+]], [[SSA]]*** [[CP0]] // CK29-DAG: store [[SSA]]** [[VAR00:%.+]], [[SSA]]*** [[CP0]]
// CK29-DAG: store i64 %{{.+}}, i64* [[S0]] // CK29-DAG: store i64 %{{.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE00]], i32 0, i32 0),
// CK29-DAG: [[VAR0]] = load [[SSB]]*, [[SSB]]** % // CK29-DAG: [[VAR0]] = load [[SSB]]*, [[SSB]]** %
// CK29-DAG: [[VAR00]] = getelementptr inbounds [[SSB]], [[SSB]]* [[VAR0]], i32 0, i32 0 // CK29-DAG: [[VAR00]] = getelementptr inbounds [[SSB]], [[SSB]]* [[VAR0]], i32 0, i32 0
// CK29-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK29-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK29-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK29-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK29-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK29-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to double**** // CK29-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to double****
// CK29-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double** // CK29-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double**
// CK29-DAG: store double*** [[VAR1:%.+]], double**** [[CBP2]] // CK29-DAG: store double*** [[VAR1:%.+]], double**** [[CBP2]]
// CK29-DAG: store double* [[VAR2:%.+]], double** [[CP2]] // CK29-DAG: store double* [[VAR2:%.+]], double** [[CP2]]
// CK29-DAG: store i64 80, i64* [[S2]]
// CK29-DAG: [[VAR1]] = getelementptr inbounds [[SSA]], [[SSA]]* %{{.+}}, i32 0, i32 1 // CK29-DAG: [[VAR1]] = getelementptr inbounds [[SSA]], [[SSA]]* %{{.+}}, i32 0, i32 1
// CK29-DAG: [[VAR2]] = getelementptr inbounds double, double* [[VAR22:%.+]], i{{.+}} 0 // CK29-DAG: [[VAR2]] = getelementptr inbounds double, double* [[VAR22:%.+]], i{{.+}} 0
// CK29-DAG: [[VAR22]] = load double*, double** %{{.+}}, // CK29-DAG: [[VAR22]] = load double*, double** %{{.+}},
// CK29: call void [[CALL00:@.+]]([[SSB]]* {{[^,]+}}) // CK29: call void [[CALL00:@.+]]([[SSB]]* {{[^,]+}})
#pragma omp target map(p->pr[:10]) #pragma omp target map(p->pr[:10])
{ {
p->pr++; p->pr++;
} }
// Region 01 // Region 01
// CK29-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE01]]{{.+}}, i8** null, i8** null) // CK29-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[SIZE01]]{{.+}}, {{.+}}getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE01]]{{.+}}, i8** null, i8** null)
// CK29-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK29-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK29-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK29-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK29-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK29-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK29-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK29-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK29-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK29-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK29-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SSB]]** // CK29-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SSB]]**
// CK29-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SSA]]**** // CK29-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SSA]]****
// CK29-DAG: store [[SSB]]* [[VAR0]], [[SSB]]** [[CBP0]] // CK29-DAG: store [[SSB]]* [[VAR0]], [[SSB]]** [[CBP0]]
// CK29-DAG: store [[SSA]]*** [[VAR000:%.+]], [[SSA]]**** [[CP0]] // CK29-DAG: store [[SSA]]*** [[VAR000:%.+]], [[SSA]]**** [[CP0]]
// CK29-DAG: store i64 %{{.+}}, i64* [[S0]] // CK29-DAG: store i64 %{{.+}}, i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE01]], i32 0, i32 0),
// CK29-DAG: [[VAR000]] = getelementptr inbounds [[SSB]], [[SSB]]* [[VAR0]], i32 0, i32 1 // CK29-DAG: [[VAR000]] = getelementptr inbounds [[SSB]], [[SSB]]* [[VAR0]], i32 0, i32 1
// CK29-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK29-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK29-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK29-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK29-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK29-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SSA]]**** // CK29-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[SSA]]****
// CK29-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double*** // CK29-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double***
// CK29-DAG: store [[SSA]]*** [[VAR000]], [[SSA]]**** [[CBP1]] // CK29-DAG: store [[SSA]]*** [[VAR000]], [[SSA]]**** [[CBP1]]
// CK29-DAG: store double** [[VAR1:%.+]], double*** [[CP1]] // CK29-DAG: store double** [[VAR1:%.+]], double*** [[CP1]]
// CK29-DAG: store i64 {{8|4}}, i64* [[S1]]
// CK29-DAG: [[VAR1]] = getelementptr inbounds [[SSA]], [[SSA]]* %{{.+}}, i32 0, i32 0 // CK29-DAG: [[VAR1]] = getelementptr inbounds [[SSA]], [[SSA]]* %{{.+}}, i32 0, i32 0
// CK29-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK29-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK29-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK29-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK29-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK29-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to double*** // CK29-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to double***
// CK29-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double** // CK29-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double**
// CK29-DAG: store double** [[VAR1]], double*** [[CBP2]] // CK29-DAG: store double** [[VAR1]], double*** [[CBP2]]
// CK29-DAG: store double* [[VAR2:%.+]], double** [[CP2]] // CK29-DAG: store double* [[VAR2:%.+]], double** [[CP2]]
// CK29-DAG: store i64 80, i64* [[S2]]
// CK29-DAG: [[VAR2]] = getelementptr inbounds double, double* [[VAR22:%.+]], i{{.+}} 0 // CK29-DAG: [[VAR2]] = getelementptr inbounds double, double* [[VAR22:%.+]], i{{.+}} 0
// CK29-DAG: [[VAR22]] = load double*, double** %{{.+}}, // CK29-DAG: [[VAR22]] = load double*, double** %{{.+}},
// CK29: call void [[CALL00:@.+]]([[SSB]]* {{[^,]+}}) // CK29: call void [[CALL00:@.+]]([[SSB]]* {{[^,]+}})
#pragma omp target map(pr->p[:10]) #pragma omp target map(pr->p[:10])
{ {
pr->p++; pr->p++;
} }
// Region 02 // Region 02
// CK29-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE02]]{{.+}}, i8** null, i8** null) // CK29-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE02]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE02]]{{.+}}, i8** null, i8** null)
// CK29-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK29-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK29-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK29-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK29-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK29-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK29-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK29-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK29-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK29-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK29-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SSB]]** // CK29-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[SSB]]**
// CK29-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SSA]]**** // CK29-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to [[SSA]]****
// CK29-DAG: store [[SSB]]* [[VAR0]], [[SSB]]** [[CBP0]] // CK29-DAG: store [[SSB]]* [[VAR0]], [[SSB]]** [[CBP0]]
// CK29-DAG: store [[SSA]]*** [[VAR000:%.+]], [[SSA]]**** [[CP0]] // CK29-DAG: store [[SSA]]*** [[VAR000:%.+]], [[SSA]]**** [[CP0]]
// CK29-DAG: store i64 %{{.+}}, i64* [[S0]] // CK29-DAG: store i64 %{{.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE02]], i32 0, i32 0),
// CK29-DAG: [[VAR000]] = getelementptr inbounds [[SSB]], [[SSB]]* [[VAR0]], i32 0, i32 1 // CK29-DAG: [[VAR000]] = getelementptr inbounds [[SSB]], [[SSB]]* [[VAR0]], i32 0, i32 1
// CK29-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK29-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK29-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK29-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK29-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK29-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to double**** // CK29-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to double****
// CK29-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double** // CK29-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to double**
// CK29-DAG: store double*** [[VAR1:%.+]], double**** [[CBP2]] // CK29-DAG: store double*** [[VAR1:%.+]], double**** [[CBP2]]
// CK29-DAG: store double* [[VAR2:%.+]], double** [[CP2]] // CK29-DAG: store double* [[VAR2:%.+]], double** [[CP2]]
// CK29-DAG: store i64 80, i64* [[S2]]
// CK29-DAG: [[VAR1]] = getelementptr inbounds [[SSA]], [[SSA]]* %{{.+}}, i32 0, i32 1 // CK29-DAG: [[VAR1]] = getelementptr inbounds [[SSA]], [[SSA]]* %{{.+}}, i32 0, i32 1
// CK29-DAG: [[VAR2]] = getelementptr inbounds double, double* [[VAR22:%.+]], i{{.+}} 0 // CK29-DAG: [[VAR2]] = getelementptr inbounds double, double* [[VAR22:%.+]], i{{.+}} 0
// CK29-DAG: [[VAR22]] = load double*, double** %{{.+}}, // CK29-DAG: [[VAR22]] = load double*, double** %{{.+}},
// CK29: call void [[CALL00:@.+]]([[SSB]]* {{[^,]+}}) // CK29: call void [[CALL00:@.+]]([[SSB]]* {{[^,]+}})
#pragma omp target map(pr->pr[:10]) #pragma omp target map(pr->pr[:10])
{ {
pr->pr++; pr->pr++;
Show All 12 Lines

clang/test/OpenMP/target_map_codegen_29.cpp

Show All 31 Lines
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY30 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY30 %s
// SIMD-ONLY30-NOT: {{__kmpc|__tgt}} // SIMD-ONLY30-NOT: {{__kmpc|__tgt}}
#ifdef CK30 #ifdef CK30
// CK30-DAG: [[BASE:%.+]] = type { i32*, i32, i32* } // CK30-DAG: [[BASE:%.+]] = type { i32*, i32, i32* }
// CK30-DAG: [[STRUCT:%.+]] = type { [[BASE]], i32*, i32*, i32, i32* } // CK30-DAG: [[STRUCT:%.+]] = type { [[BASE]], i32*, i32*, i32, i32* }
// CK30-LABEL: @.__omp_offloading_{{.*}}map_with_deep_copy{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK30-LABEL: @.__omp_offloading_{{.*}}map_with_deep_copy{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK30: [[SIZE00:@.+]] = private unnamed_addr global [4 x i64] [i64 0, i64 {{56|28}}, i64 4, i64 4]
// The first element: 0x20 - OMP_MAP_TARGET_PARAM // The first element: 0x20 - OMP_MAP_TARGET_PARAM
// 2: 0x1000000000003 - OMP_MAP_MEMBER_OF(0) | OMP_MAP_TO | OMP_MAP_FROM - copies all the data in structs excluding deep-copied elements (from &s to end of s). // 2: 0x1000000000003 - OMP_MAP_MEMBER_OF(0) | OMP_MAP_TO | OMP_MAP_FROM - copies all the data in structs excluding deep-copied elements (from &s to end of s).
// 3-4: 0x1000000000013 - OMP_MAP_MEMBER_OF(0) | OMP_MAP_PTR_AND_OBJ | OMP_MAP_TO | OMP_MAP_FROM - deep copy of the pointers + pointee. // 3-4: 0x1000000000013 - OMP_MAP_MEMBER_OF(0) | OMP_MAP_PTR_AND_OBJ | OMP_MAP_TO | OMP_MAP_FROM - deep copy of the pointers + pointee.
// CK30: [[MTYPE00:@.+]] = private {{.*}}constant [4 x i64] [i64 32, i64 281474976710659, i64 281474976710675, i64 281474976710675] // CK30: [[MTYPE00:@.+]] = private {{.*}}constant [4 x i64] [i64 32, i64 281474976710659, i64 281474976710675, i64 281474976710675]
typedef struct { typedef struct {
int *ptrBase; int *ptrBase;
int valBase; int valBase;
int *ptrBase1; int *ptrBase1;
} Base; } Base;
typedef struct StructWithPtrTag : public Base { typedef struct StructWithPtrTag : public Base {
int *ptr; int *ptr;
int *ptr2; int *ptr2;
int val; int val;
int *ptr1; int *ptr1;
} StructWithPtr; } StructWithPtr;
// CK30-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @.__omp_offloading_{{.*}}map_with_deep_copy{{.*}}_l{{[0-9]+}}.region_id, i32 4, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[MTYPE00]], i32 0, i32 0), i8** null, i8** null) // CK30-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @.__omp_offloading_{{.*}}map_with_deep_copy{{.*}}_l{{[0-9]+}}.region_id, i32 4, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE00]], i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[MTYPE00]], i32 0, i32 0), i8** null, i8** null)
// CK30-DAG: [[GEPS]] = getelementptr inbounds [4 x i{{64|32}}], [4 x i64]* [[SIZES:%.+]], i32 0, i32 0
// CK30-DAG: [[GEPP]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[PTRS:%.+]], i32 0, i32 0 // CK30-DAG: [[GEPP]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[PTRS:%.+]], i32 0, i32 0
// CK30-DAG: [[GEPBP]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BASES:%.+]], i32 0, i32 0 // CK30-DAG: [[GEPBP]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BASES:%.+]], i32 0, i32 0
// CK30-DAG: [[BASE_PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BASES]], i32 0, i32 0 // CK30-DAG: [[BASE_PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BASES]], i32 0, i32 0
// CK30-DAG: [[BC:%.+]] = bitcast i8** [[BASE_PTR]] to [[STRUCT]]** // CK30-DAG: [[BC:%.+]] = bitcast i8** [[BASE_PTR]] to [[STRUCT]]**
// CK30-DAG: store [[STRUCT]]* [[S:%.+]], [[STRUCT]]** [[BC]], // CK30-DAG: store [[STRUCT]]* [[S:%.+]], [[STRUCT]]** [[BC]],
// CK30-DAG: [[PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[PTRS]], i32 0, i32 0 // CK30-DAG: [[PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[PTRS]], i32 0, i32 0
// CK30-DAG: [[BC:%.+]] = bitcast i8** [[PTR]] to [[STRUCT]]** // CK30-DAG: [[BC:%.+]] = bitcast i8** [[PTR]] to [[STRUCT]]**
// CK30-DAG: store [[STRUCT]]* [[S]], [[STRUCT]]** [[BC]], // CK30-DAG: store [[STRUCT]]* [[S]], [[STRUCT]]** [[BC]],
// CK30-DAG: [[SIZE:%.+]] = getelementptr inbounds [4 x i{{64|32}}], [4 x i64]* [[SIZES]], i32 0, i32 0 // CK30-DAG: store i64 [[S_ALLOC_SIZE:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE00]], i32 0, i32 0),
// CK30-DAG: store i64 [[S_ALLOC_SIZE:%.+]], i64* [[SIZE]],
// CK30-DAG: [[S_ALLOC_SIZE]] = sdiv exact i64 [[DIFF:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK30-DAG: [[S_ALLOC_SIZE]] = sdiv exact i64 [[DIFF:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK30-DAG: [[DIFF]] = sub i64 [[S_END_BC:%.+]], [[S_BEGIN_BC:%.+]] // CK30-DAG: [[DIFF]] = sub i64 [[S_END_BC:%.+]], [[S_BEGIN_BC:%.+]]
// CK30-DAG: [[S_BEGIN_BC]] = ptrtoint i8* [[S_BEGIN:%.+]] to i64 // CK30-DAG: [[S_BEGIN_BC]] = ptrtoint i8* [[S_BEGIN:%.+]] to i64
// CK30-DAG: [[S_END_BC]] = ptrtoint i8* [[S_END:%.+]] to i64 // CK30-DAG: [[S_END_BC]] = ptrtoint i8* [[S_END:%.+]] to i64
// CK30-DAG: [[S_BEGIN]] = bitcast [[STRUCT]]* [[S]] to i8* // CK30-DAG: [[S_BEGIN]] = bitcast [[STRUCT]]* [[S]] to i8*
// CK30-DAG: [[S_END]] = bitcast [[STRUCT]]* [[REAL_S_END:%.+]] to i8* // CK30-DAG: [[S_END]] = bitcast [[STRUCT]]* [[REAL_S_END:%.+]] to i8*
// CK30-DAG: [[REAL_S_END]] = getelementptr [[STRUCT]], [[STRUCT]]* [[S]], i32 1 // CK30-DAG: [[REAL_S_END]] = getelementptr [[STRUCT]], [[STRUCT]]* [[S]], i32 1
// CK30-DAG: [[BASE_PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BASES]], i32 0, i32 1 // CK30-DAG: [[BASE_PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BASES]], i32 0, i32 1
// CK30-DAG: [[BC:%.+]] = bitcast i8** [[BASE_PTR]] to [[STRUCT]]** // CK30-DAG: [[BC:%.+]] = bitcast i8** [[BASE_PTR]] to [[STRUCT]]**
// CK30-DAG: store [[STRUCT]]* [[S]], [[STRUCT]]** [[BC]], // CK30-DAG: store [[STRUCT]]* [[S]], [[STRUCT]]** [[BC]],
// CK30-DAG: [[PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[PTRS]], i32 0, i32 1 // CK30-DAG: [[PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[PTRS]], i32 0, i32 1
// CK30-DAG: [[BC:%.+]] = bitcast i8** [[PTR]] to [[STRUCT]]** // CK30-DAG: [[BC:%.+]] = bitcast i8** [[PTR]] to [[STRUCT]]**
// CK30-DAG: store [[STRUCT]]* [[S]], [[STRUCT]]** [[BC]], // CK30-DAG: store [[STRUCT]]* [[S]], [[STRUCT]]** [[BC]],
// CK30-DAG: [[SIZE:%.+]] = getelementptr inbounds [4 x i64], [4 x i64]* [[SIZES]], i32 0, i32 1
// CK30-DAG: store i64 {{56|28}}, i64* [[SIZE]],
// CK30-DAG: [[BASE_PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BASES]], i32 0, i32 2 // CK30-DAG: [[BASE_PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BASES]], i32 0, i32 2
// CK30-DAG: [[BC:%.+]] = bitcast i8** [[BASE_PTR]] to i32*** // CK30-DAG: [[BC:%.+]] = bitcast i8** [[BASE_PTR]] to i32***
// CK30-DAG: store i32** [[S_PTR1:%.+]], i32*** [[BC]], // CK30-DAG: store i32** [[S_PTR1:%.+]], i32*** [[BC]],
// CK30-DAG: [[PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[PTRS]], i32 0, i32 2 // CK30-DAG: [[PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[PTRS]], i32 0, i32 2
// CK30-DAG: [[BC:%.+]] = bitcast i8** [[PTR]] to i32** // CK30-DAG: [[BC:%.+]] = bitcast i8** [[PTR]] to i32**
// CK30-DAG: store i32* [[S_PTR1_BEGIN:%.+]], i32** [[BC]], // CK30-DAG: store i32* [[S_PTR1_BEGIN:%.+]], i32** [[BC]],
// CK30-DAG: [[SIZE:%.+]] = getelementptr inbounds [4 x i64], [4 x i64]* [[SIZES]], i32 0, i32 2
// CK30-DAG: store i64 4, i64* [[SIZE]],
// CK30-DAG: [[S_PTR1]] = getelementptr inbounds [[STRUCT]], [[STRUCT]]* [[S]], i32 0, i32 4 // CK30-DAG: [[S_PTR1]] = getelementptr inbounds [[STRUCT]], [[STRUCT]]* [[S]], i32 0, i32 4
// CK30-DAG: [[S_PTR1_BEGIN]] = getelementptr inbounds i32, i32* [[S_PTR1_BEGIN_REF:%.+]], i{{64|32}} 0 // CK30-DAG: [[S_PTR1_BEGIN]] = getelementptr inbounds i32, i32* [[S_PTR1_BEGIN_REF:%.+]], i{{64|32}} 0
// CK30-DAG: [[S_PTR1_BEGIN_REF]] = load i32*, i32** [[S_PTR1:%.+]], // CK30-DAG: [[S_PTR1_BEGIN_REF]] = load i32*, i32** [[S_PTR1:%.+]],
// CK30-DAG: [[S_PTR1]] = getelementptr inbounds [[STRUCT]], [[STRUCT]]* [[S]], i32 0, i32 4 // CK30-DAG: [[S_PTR1]] = getelementptr inbounds [[STRUCT]], [[STRUCT]]* [[S]], i32 0, i32 4
// CK30-DAG: [[BASE_PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BASES]], i32 0, i32 3 // CK30-DAG: [[BASE_PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BASES]], i32 0, i32 3
// CK30-DAG: [[BC:%.+]] = bitcast i8** [[BASE_PTR]] to i32*** // CK30-DAG: [[BC:%.+]] = bitcast i8** [[BASE_PTR]] to i32***
// CK30-DAG: store i32** [[S_PTRBASE1:%.+]], i32*** [[BC]], // CK30-DAG: store i32** [[S_PTRBASE1:%.+]], i32*** [[BC]],
// CK30-DAG: [[PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[PTRS]], i32 0, i32 3 // CK30-DAG: [[PTR:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[PTRS]], i32 0, i32 3
// CK30-DAG: [[BC:%.+]] = bitcast i8** [[PTR]] to i32** // CK30-DAG: [[BC:%.+]] = bitcast i8** [[PTR]] to i32**
// CK30-DAG: store i32* [[S_PTRBASE1_BEGIN:%.+]], i32** [[BC]], // CK30-DAG: store i32* [[S_PTRBASE1_BEGIN:%.+]], i32** [[BC]],
// CK30-DAG: [[SIZE:%.+]] = getelementptr inbounds [4 x i{{64|32}}], [4 x i{{64|32}}]* [[SIZES]], i32 0, i32 3
// CK30-DAG: store i{{64|32}} 4, i{{64|32}}* [[SIZE]],
// CK30-DAG: [[S_PTRBASE1]] = getelementptr inbounds [[BASE]], [[BASE]]* [[S_BASE:%.+]], i32 0, i32 2 // CK30-DAG: [[S_PTRBASE1]] = getelementptr inbounds [[BASE]], [[BASE]]* [[S_BASE:%.+]], i32 0, i32 2
// CK30-DAG: [[S_BASE]] = bitcast [[STRUCT]]* [[S]] to [[BASE]]* // CK30-DAG: [[S_BASE]] = bitcast [[STRUCT]]* [[S]] to [[BASE]]*
// CK30-DAG: [[S_PTRBASE1_BEGIN]] = getelementptr inbounds i32, i32* [[S_PTRBASE1_BEGIN_REF:%.+]], i{{64|32}} 0 // CK30-DAG: [[S_PTRBASE1_BEGIN]] = getelementptr inbounds i32, i32* [[S_PTRBASE1_BEGIN_REF:%.+]], i{{64|32}} 0
// CK30-DAG: [[S_PTRBASE1_BEGIN_REF]] = load i32*, i32** [[S_PTRBASE1:%.+]], // CK30-DAG: [[S_PTRBASE1_BEGIN_REF]] = load i32*, i32** [[S_PTRBASE1:%.+]],
// CK30-DAG: [[S_PTRBASE1]] = getelementptr inbounds [[BASE]], [[BASE]]* [[S_BASE:%.+]], i32 0, i32 2 // CK30-DAG: [[S_PTRBASE1]] = getelementptr inbounds [[BASE]], [[BASE]]* [[S_BASE:%.+]], i32 0, i32 2
// CK30-DAG: [[S_BASE]] = bitcast [[STRUCT]]* [[S]] to [[BASE]]* // CK30-DAG: [[S_BASE]] = bitcast [[STRUCT]]* [[S]] to [[BASE]]*
void map_with_deep_copy() { void map_with_deep_copy() {
StructWithPtr s; StructWithPtr s;
Show All 10 Lines

clang/test/OpenMP/target_map_codegen_31.cpp

Show All 35 Lines
// CK31A: [[ST:%.+]] = type { i32, i32 } // CK31A: [[ST:%.+]] = type { i32, i32 }
struct ST { struct ST {
int i; int i;
int j; int j;
}; };
// CK31A-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK31A-LABEL: @.__omp_offloading_{{.*}}explicit_maps_single{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// //
// CK31A: [[SIZE00:@.+]] = private {{.*}}global [7 x i64] [i64 0, i64 4, i64 4, i64 4, i64 0, i64 4, i64 4]
// PRESENT=0x1000 | TARGET_PARAM=0x20 = 0x1020 // PRESENT=0x1000 | TARGET_PARAM=0x20 = 0x1020
// CK31A-USE: [[MTYPE00:@.+]] = private {{.*}}constant [7 x i64] [i64 [[#0x1020]], // CK31A-USE: [[MTYPE00:@.+]] = private {{.*}}constant [7 x i64] [i64 [[#0x1020]],
// CK31A-NOUSE: [[MTYPE00:@.+]] = private {{.*}}constant [7 x i64] [i64 [[#0x1000]], // CK31A-NOUSE: [[MTYPE00:@.+]] = private {{.*}}constant [7 x i64] [i64 [[#0x1000]],
// //
// MEMBER_OF_1=0x1000000000000 | PRESENT=0x1000 | FROM=0x2 | TO=0x1 = 0x1000000001003 // MEMBER_OF_1=0x1000000000000 | PRESENT=0x1000 | FROM=0x2 | TO=0x1 = 0x1000000001003
// MEMBER_OF_1=0x1000000000000 | FROM=0x2 | TO=0x1 = 0x1000000000003 // MEMBER_OF_1=0x1000000000000 | FROM=0x2 | TO=0x1 = 0x1000000000003
// PRESENT=0x1000 | TARGET_PARAM=0x20 | FROM=0x2 | TO=0x1 = 0x1023 // PRESENT=0x1000 | TARGET_PARAM=0x20 | FROM=0x2 | TO=0x1 = 0x1023
// CK31A-USE-SAME: {{^}} i64 [[#0x1000000001003]], i64 [[#0x1000000000003]], i64 [[#0x1023]], // CK31A-USE-SAME: {{^}} i64 [[#0x1000000001003]], i64 [[#0x1000000000003]], i64 [[#0x1023]],
Show All 27 Lines
// Make sure the struct picks up present even if another element of the struct // Make sure the struct picks up present even if another element of the struct
// doesn't have present. // doesn't have present.
// Region 00 // Region 00
// CK31A: [[ST1_J:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[ST1]], i{{.+}} 0, i{{.+}} 1 // CK31A: [[ST1_J:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[ST1]], i{{.+}} 0, i{{.+}} 1
// CK31A: [[ST1_I:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[ST1]], i{{.+}} 0, i{{.+}} 0 // CK31A: [[ST1_I:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[ST1]], i{{.+}} 0, i{{.+}} 0
// CK31A: [[ST2_I:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[ST2]], i{{.+}} 0, i{{.+}} 0 // CK31A: [[ST2_I:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[ST2]], i{{.+}} 0, i{{.+}} 0
// CK31A: [[ST2_J:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[ST2]], i{{.+}} 0, i{{.+}} 1 // CK31A: [[ST2_J:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[ST2]], i{{.+}} 0, i{{.+}} 1
// CK31A-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 7, i8** [[GEPBP:%[0-9]+]], i8** [[GEPP:%[0-9]+]], i64* [[GEPS:%.+]], i64* getelementptr {{.+}}[7 x i{{.+}}]* [[MTYPE00]]{{.+}}) // CK31A-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 7, i8** [[GEPBP:%[0-9]+]], i8** [[GEPP:%[0-9]+]], i64* getelementptr {{.+}}[7 x i{{.+}}]* [[SIZE00]]{{.+}}, i64* getelementptr {{.+}}[7 x i{{.+}}]* [[MTYPE00]]{{.+}})
// CK31A-DAG: [[GEPS]] = getelementptr inbounds [7 x i64], [7 x i64]* [[S:%.+]], i{{.+}} 0, i{{.+}} 0
// CK31A-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK31A-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK31A-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK31A-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// st1 // st1
// CK31A-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK31A-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK31A-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK31A-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK31A-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK31A-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK31A-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK31A-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK31A-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK31A-DAG: store [[ST]]* [[ST1]], [[ST]]** [[CBP0]] // CK31A-DAG: store [[ST]]* [[ST1]], [[ST]]** [[CBP0]]
// CK31A-DAG: store i32* [[ST1_I]], i32** [[CP0]] // CK31A-DAG: store i32* [[ST1_I]], i32** [[CP0]]
// CK31A-DAG: store i64 %{{.+}}, i64* [[S0]] // CK31A-DAG: store i64 %{{.+}}, i64* getelementptr inbounds ([7 x i64], [7 x i64]* [[SIZE00]], i32 0, i32 0),
// st1.j // st1.j
// CK31A-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK31A-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK31A-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK31A-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK31A-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK31A-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]** // CK31A-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]**
// CK31A-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK31A-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK31A-DAG: store [[ST]]* [[ST1]], [[ST]]** [[CBP1]] // CK31A-DAG: store [[ST]]* [[ST1]], [[ST]]** [[CBP1]]
// CK31A-DAG: store i32* [[ST1_J]], i32** [[CP1]] // CK31A-DAG: store i32* [[ST1_J]], i32** [[CP1]]
// CK31A-DAG: store i64 4, i64* [[S1]]
// st1.i // st1.i
// CK31A-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK31A-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK31A-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK31A-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK31A-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK31A-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]** // CK31A-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]**
// CK31A-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32** // CK31A-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32**
// CK31A-DAG: store [[ST]]* [[ST1]], [[ST]]** [[CBP2]] // CK31A-DAG: store [[ST]]* [[ST1]], [[ST]]** [[CBP2]]
// CK31A-DAG: store i32* [[ST1_I]], i32** [[CP2]] // CK31A-DAG: store i32* [[ST1_I]], i32** [[CP2]]
// CK31A-DAG: store i64 4, i64* [[S2]]
// a // a
// CK31A-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3 // CK31A-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3
// CK31A-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3 // CK31A-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3
// CK31A-DAG: [[S3:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 3
// CK31A-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to i32** // CK31A-DAG: [[CBP3:%.+]] = bitcast i8** [[BP3]] to i32**
// CK31A-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to i32** // CK31A-DAG: [[CP3:%.+]] = bitcast i8** [[P3]] to i32**
// CK31A-DAG: store i32* [[A]], i32** [[CBP3]] // CK31A-DAG: store i32* [[A]], i32** [[CBP3]]
// CK31A-DAG: store i32* [[A]], i32** [[CP3]] // CK31A-DAG: store i32* [[A]], i32** [[CP3]]
// CK31A-DAG: store i64 4, i64* [[S3]]
// st2 // st2
// CK31A-DAG: [[BP4:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 4 // CK31A-DAG: [[BP4:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 4
// CK31A-DAG: [[P4:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 4 // CK31A-DAG: [[P4:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 4
// CK31A-DAG: [[S4:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 4
// CK31A-DAG: [[CBP4:%.+]] = bitcast i8** [[BP4]] to [[ST]]** // CK31A-DAG: [[CBP4:%.+]] = bitcast i8** [[BP4]] to [[ST]]**
// CK31A-DAG: [[CP4:%.+]] = bitcast i8** [[P4]] to i32** // CK31A-DAG: [[CP4:%.+]] = bitcast i8** [[P4]] to i32**
// CK31A-DAG: store [[ST]]* [[ST2]], [[ST]]** [[CBP4]] // CK31A-DAG: store [[ST]]* [[ST2]], [[ST]]** [[CBP4]]
// CK31A-DAG: store i32* [[ST2_I]], i32** [[CP4]] // CK31A-DAG: store i32* [[ST2_I]], i32** [[CP4]]
// CK31A-DAG: store i64 %{{.+}}, i64* [[S4]] // CK31A-DAG: store i64 %{{.+}}, i64* getelementptr inbounds ([7 x i64], [7 x i64]* [[SIZE00]], i32 0, i32 4),
// st2.i // st2.i
// CK31A-DAG: [[BP5:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 5 // CK31A-DAG: [[BP5:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 5
// CK31A-DAG: [[P5:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 5 // CK31A-DAG: [[P5:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 5
// CK31A-DAG: [[S5:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 5
// CK31A-DAG: [[CBP5:%.+]] = bitcast i8** [[BP5]] to [[ST]]** // CK31A-DAG: [[CBP5:%.+]] = bitcast i8** [[BP5]] to [[ST]]**
// CK31A-DAG: [[CP5:%.+]] = bitcast i8** [[P5]] to i32** // CK31A-DAG: [[CP5:%.+]] = bitcast i8** [[P5]] to i32**
// CK31A-DAG: store [[ST]]* [[ST2]], [[ST]]** [[CBP5]] // CK31A-DAG: store [[ST]]* [[ST2]], [[ST]]** [[CBP5]]
// CK31A-DAG: store i32* [[ST2_I]], i32** [[CP5]] // CK31A-DAG: store i32* [[ST2_I]], i32** [[CP5]]
// CK31A-DAG: store i64 4, i64* [[S5]]
// st2.j // st2.j
// CK31A-DAG: [[BP6:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 6 // CK31A-DAG: [[BP6:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 6
// CK31A-DAG: [[P6:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 6 // CK31A-DAG: [[P6:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 6
// CK31A-DAG: [[S6:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 6
// CK31A-DAG: [[CBP6:%.+]] = bitcast i8** [[BP6]] to [[ST]]** // CK31A-DAG: [[CBP6:%.+]] = bitcast i8** [[BP6]] to [[ST]]**
// CK31A-DAG: [[CP6:%.+]] = bitcast i8** [[P6]] to i32** // CK31A-DAG: [[CP6:%.+]] = bitcast i8** [[P6]] to i32**
// CK31A-DAG: store [[ST]]* [[ST2]], [[ST]]** [[CBP6]] // CK31A-DAG: store [[ST]]* [[ST2]], [[ST]]** [[CBP6]]
// CK31A-DAG: store i32* [[ST2_J]], i32** [[CP6]] // CK31A-DAG: store i32* [[ST2_J]], i32** [[CP6]]
// CK31A-DAG: store i64 4, i64* [[S6]]
// CK31A-USE: call void [[CALL00:@.+]]([[ST]]* [[ST1]], i32* [[A]], [[ST]]* [[ST2]]) // CK31A-USE: call void [[CALL00:@.+]]([[ST]]* [[ST1]], i32* [[A]], [[ST]]* [[ST2]])
// CK31A-NOUSE: call void [[CALL00:@.+]]() // CK31A-NOUSE: call void [[CALL00:@.+]]()
#pragma omp target map(tofrom \ #pragma omp target map(tofrom \
: st1.i) map(present, tofrom \ : st1.i) map(present, tofrom \
: a, st1.j, st2.i) map(tofrom \ : a, st1.j, st2.i) map(tofrom \
: st2.j) : st2.j)
{ {
Show All 36 Lines

clang/test/OpenMP/target_map_codegen_32.cpp

Show All 34 Lines
// CK31B: [[ST:%.+]] = type { i32, i32 } // CK31B: [[ST:%.+]] = type { i32, i32 }
// PRESENT=0x1000 | TARGET_PARAM=0x20 = 0x1020 // PRESENT=0x1000 | TARGET_PARAM=0x20 = 0x1020
// MEMBER_OF_1=0x1000000000000 | PRESENT=0x1000 | FROM=0x2 | TO=0x1 = 0x1000000001003 // MEMBER_OF_1=0x1000000000000 | PRESENT=0x1000 | FROM=0x2 | TO=0x1 = 0x1000000001003
// MEMBER_OF_1=0x1000000000000 | FROM=0x2 | TO=0x1 = 0x1000000000003 // MEMBER_OF_1=0x1000000000000 | FROM=0x2 | TO=0x1 = 0x1000000000003
// CK31B-LABEL: @.__omp_offloading_{{.*}}test_present_members{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0 // CK31B-LABEL: @.__omp_offloading_{{.*}}test_present_members{{.*}}_l{{[0-9]+}}.region_id = weak constant i8 0
// CK31B: [[SIZE00:@.+]] = private {{.*}}global [3 x i64] [i64 0, i64 4, i64 4]
// CK31B-USE: [[MTYPE00:@.+]] = private {{.*}}constant [3 x i64] [i64 [[#0x1020]], // CK31B-USE: [[MTYPE00:@.+]] = private {{.*}}constant [3 x i64] [i64 [[#0x1020]],
// CK31B-NOUSE: [[MTYPE00:@.+]] = private {{.*}}constant [3 x i64] [i64 [[#0x1000]], // CK31B-NOUSE: [[MTYPE00:@.+]] = private {{.*}}constant [3 x i64] [i64 [[#0x1000]],
// CK31B-USE-SAME: {{^}} i64 [[#0x1000000001003]], i64 [[#0x1000000000003]]] // CK31B-USE-SAME: {{^}} i64 [[#0x1000000001003]], i64 [[#0x1000000000003]]]
// CK31B-NOUSE-SAME: {{^}} i64 [[#0x1000000001003]], i64 [[#0x1000000000003]]] // CK31B-NOUSE-SAME: {{^}} i64 [[#0x1000000001003]], i64 [[#0x1000000000003]]]
struct ST { struct ST {
int i; int i;
int j; int j;
// CK31B-LABEL: define {{.*}}test_present_members{{.*}}( // CK31B-LABEL: define {{.*}}test_present_members{{.*}}(
void test_present_members() { void test_present_members() {
// Make sure the struct picks up present even if another element of the // Make sure the struct picks up present even if another element of the
// struct doesn't have present. // struct doesn't have present.
// Region 00 // Region 00
// CK31B: [[J:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[THIS:%.+]], i{{.+}} 0, i{{.+}} 1 // CK31B: [[J:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[THIS:%.+]], i{{.+}} 0, i{{.+}} 1
// CK31B: [[I:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[THIS]], i{{.+}} 0, i{{.+}} 0 // CK31B: [[I:%.+]] = getelementptr inbounds [[ST]], [[ST]]* [[THIS]], i{{.+}} 0, i{{.+}} 0
// CK31B-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%[0-9]+]], i8** [[GEPP:%[0-9]+]], i64* [[GEPS:%.+]], i64* getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE00]]{{.+}}) // CK31B-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 {{[^,]+}}, i8* {{[^,]+}}, i32 3, i8** [[GEPBP:%[0-9]+]], i8** [[GEPP:%[0-9]+]], i64* getelementptr {{.+}}[3 x i{{.+}}]* [[SIZE00]]{{.+}}, i64* getelementptr {{.+}}[3 x i{{.+}}]* [[MTYPE00]]{{.+}})
// CK31B-DAG: [[GEPS]] = getelementptr inbounds [3 x i64], [3 x i64]* [[S:%.+]], i{{.+}} 0, i{{.+}} 0
// CK31B-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK31B-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK31B-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK31B-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// this // this
// CK31B-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK31B-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK31B-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK31B-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK31B-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK31B-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK31B-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK31B-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32** // CK31B-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to i32**
// CK31B-DAG: store [[ST]]* [[THIS]], [[ST]]** [[CBP0]] // CK31B-DAG: store [[ST]]* [[THIS]], [[ST]]** [[CBP0]]
// CK31B-DAG: store i32* [[I]], i32** [[CP0]] // CK31B-DAG: store i32* [[I]], i32** [[CP0]]
// CK31B-DAG: store i64 %{{.+}}, i64* [[S0]] // CK31B-DAG: store i64 %{{.+}}, i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE00]], i32 0, i32 0),
// j // j
// CK31B-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK31B-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK31B-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK31B-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK31B-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK31B-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]** // CK31B-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to [[ST]]**
// CK31B-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32** // CK31B-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to i32**
// CK31B-DAG: store [[ST]]* [[THIS]], [[ST]]** [[CBP1]] // CK31B-DAG: store [[ST]]* [[THIS]], [[ST]]** [[CBP1]]
// CK31B-DAG: store i32* [[J]], i32** [[CP1]] // CK31B-DAG: store i32* [[J]], i32** [[CP1]]
// CK31B-DAG: store i64 4, i64* [[S1]]
// i // i
// CK31B-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK31B-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK31B-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK31B-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK31B-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK31B-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]** // CK31B-DAG: [[CBP2:%.+]] = bitcast i8** [[BP2]] to [[ST]]**
// CK31B-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32** // CK31B-DAG: [[CP2:%.+]] = bitcast i8** [[P2]] to i32**
// CK31B-DAG: store [[ST]]* [[THIS]], [[ST]]** [[CBP2]] // CK31B-DAG: store [[ST]]* [[THIS]], [[ST]]** [[CBP2]]
// CK31B-DAG: store i32* [[I]], i32** [[CP2]] // CK31B-DAG: store i32* [[I]], i32** [[CP2]]
// CK31B-DAG: store i64 4, i64* [[S2]]
// CK31B-USE: call void [[CALL00:@.+]]([[ST]]* [[THIS]]) // CK31B-USE: call void [[CALL00:@.+]]([[ST]]* [[THIS]])
// CK31B-NOUSE: call void [[CALL00:@.+]]() // CK31B-NOUSE: call void [[CALL00:@.+]]()
#pragma omp target map(tofrom \ #pragma omp target map(tofrom \
: i) map(present, tofrom \ : i) map(present, tofrom \
: j) : j)
{ {
#ifdef USE #ifdef USE
Show All 16 Lines

clang/test/OpenMP/target_map_codegen_34.cpp

Show All 29 Lines
class S { class S {
int a; int a;
C c; C c;
int b; int b;
public: public:
void foo(); void foo();
}; };
// CK34-DAG: [[SIZE_TO:@.+]] = private {{.*}}global [4 x i64] [i64 0, i64 0, i64 0, i64 {{16|8}}]
// TARGET_PARAM = 0x20 // TARGET_PARAM = 0x20
// MEMBER_OF_1 | TO = 0x1000000000001 // MEMBER_OF_1 | TO = 0x1000000000001
// MEMBER_OF_1 | IMPLICIT | TO = 0x1000000000201 // MEMBER_OF_1 | IMPLICIT | TO = 0x1000000000201
// CK34-DAG: [[MTYPE_TO:@.+]] = {{.+}}constant [4 x i64] [i64 [[#0x20]], i64 [[#0x1000000000001]], i64 [[#0x1000000000001]], i64 [[#0x1000000000201]]] // CK34-DAG: [[MTYPE_TO:@.+]] = {{.+}}constant [4 x i64] [i64 [[#0x20]], i64 [[#0x1000000000001]], i64 [[#0x1000000000001]], i64 [[#0x1000000000201]]]
// CK34-DAG: [[SIZE_FROM:@.+]] = private {{.*}}global [4 x i64] [i64 0, i64 0, i64 0, i64 {{16|8}}]
// TARGET_PARAM = 0x20 // TARGET_PARAM = 0x20
// MEMBER_OF_1 | FROM = 0x1000000000002 // MEMBER_OF_1 | FROM = 0x1000000000002
// MEMBER_OF_1 | IMPLICIT | FROM = 0x1000000000202 // MEMBER_OF_1 | IMPLICIT | FROM = 0x1000000000202
// CK34-DAG: [[MTYPE_FROM:@.+]] = {{.+}}constant [4 x i64] [i64 [[#0x20]], i64 [[#0x1000000000002]], i64 [[#0x1000000000002]], i64 [[#0x1000000000202]]] // CK34-DAG: [[MTYPE_FROM:@.+]] = {{.+}}constant [4 x i64] [i64 [[#0x20]], i64 [[#0x1000000000002]], i64 [[#0x1000000000002]], i64 [[#0x1000000000202]]]
void default_mapper() { void default_mapper() {
S s; S s;
// CK34-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{.+}}, i32 4, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[MTYPE_TO]]{{.+}}, i8** null, i8** [[GEPMF:%.+]]) // CK34-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{.+}}, i32 4, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[SIZE_TO]]{{.+}}, {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[MTYPE_TO]]{{.+}}, i8** null, i8** [[GEPMF:%.+]])
// CK34-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK34-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK34-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK34-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK34-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK34-DAG: [[GEPMF]] = bitcast [4 x i8*]* [[MF:%.+]] to i8** // CK34-DAG: [[GEPMF]] = bitcast [4 x i8*]* [[MF:%.+]] to i8**
// pass TARGET_PARAM {&s, &s, ((void*)(&s+1)-(void*)&s)} // pass TARGET_PARAM {&s, &s, ((void*)(&s+1)-(void*)&s)}
// CK34-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK34-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK34-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK34-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK34-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK34-DAG: [[MF0:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 0 // CK34-DAG: [[MF0:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 0
// CK34-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to %class.S** // CK34-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to %class.S**
// CK34-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to %class.S** // CK34-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to %class.S**
// CK34-DAG: store %class.S* [[S_ADDR:%.+]], %class.S** [[BPC0]], // CK34-DAG: store %class.S* [[S_ADDR:%.+]], %class.S** [[BPC0]],
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC0]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC0]],
// CK34-DAG: store i64 [[S_SIZE:%.+]], i64* [[S0]], // CK34-DAG: store i64 [[S_SIZE:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE_TO]], i32 0, i32 0),
// CK34-DAG: store i8* null, i8** [[MF0]], // CK34-DAG: store i8* null, i8** [[MF0]],
// CK34-DAG: [[S_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK34-DAG: [[S_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK34-DAG: [[SZ]] = sub i64 [[S_1_INTPTR:%.+]], [[S_INTPTR:%.+]] // CK34-DAG: [[SZ]] = sub i64 [[S_1_INTPTR:%.+]], [[S_INTPTR:%.+]]
// CK34-DAG: [[S_1_INTPTR]] = ptrtoint i8* [[S_1_VOID:%.+]] to i64 // CK34-DAG: [[S_1_INTPTR]] = ptrtoint i8* [[S_1_VOID:%.+]] to i64
// CK34-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64 // CK34-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64
// CK34-DAG: [[S_1_VOID]] = bitcast %class.S* [[S_1:%.+]] to i8* // CK34-DAG: [[S_1_VOID]] = bitcast %class.S* [[S_1:%.+]] to i8*
// CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8* // CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8*
// CK34-DAG: [[S_1]] = getelementptr %class.S, %class.S* [[S_ADDR]], i32 1 // CK34-DAG: [[S_1]] = getelementptr %class.S, %class.S* [[S_ADDR]], i32 1
// pass MEMBER_OF_1 | TO {&s, &s, ((void*)(&s.a+1)-(void*)&s)} to copy the data of s.a. // pass MEMBER_OF_1 | TO {&s, &s, ((void*)(&s.a+1)-(void*)&s)} to copy the data of s.a.
// CK34-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK34-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK34-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK34-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK34-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK34-DAG: [[MF1:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 1 // CK34-DAG: [[MF1:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 1
// CK34-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to %class.S** // CK34-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to %class.S**
// CK34-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to %class.S** // CK34-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to %class.S**
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC1]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC1]],
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC1]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC1]],
// CK34-DAG: store i64 [[A_SIZE:%.+]], i64* [[S1]], // CK34-DAG: store i64 [[A_SIZE:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE_TO]], i32 0, i32 1),
// CK34-DAG: store i8* null, i8** [[MF1]], // CK34-DAG: store i8* null, i8** [[MF1]],
// CK34-DAG: [[A_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK34-DAG: [[A_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK34-DAG: [[SZ]] = sub i64 [[C_BEGIN_INTPTR:%.+]], [[S_INTPTR:%.+]] // CK34-DAG: [[SZ]] = sub i64 [[C_BEGIN_INTPTR:%.+]], [[S_INTPTR:%.+]]
// CK34-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64 // CK34-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64
// CK34-DAG: [[C_BEGIN_INTPTR]] = ptrtoint i8* [[C_BEGIN_VOID:%.+]] to i64 // CK34-DAG: [[C_BEGIN_INTPTR]] = ptrtoint i8* [[C_BEGIN_VOID:%.+]] to i64
// CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8* // CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8*
// CK34-DAG: [[C_BEGIN_VOID]] = bitcast %class.C* [[C_ADDR:%.+]] to i8* // CK34-DAG: [[C_BEGIN_VOID]] = bitcast %class.C* [[C_ADDR:%.+]] to i8*
// CK34-64-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 2 // CK34-64-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 2
// CK34-32-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1 // CK34-32-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1
// pass MEMBER_OF_1 | TO {&s, &s.c+1, ((void*)(&s)+31+1-(void*)(&s.c+1))} to copy the data of s.b. // pass MEMBER_OF_1 | TO {&s, &s.c+1, ((void*)(&s)+31+1-(void*)(&s.c+1))} to copy the data of s.b.
// CK34-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK34-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK34-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK34-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK34-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK34-DAG: [[MF2:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 2 // CK34-DAG: [[MF2:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 2
// CK34-DAG: [[BPC2:%.+]] = bitcast i8** [[BP2]] to %class.S** // CK34-DAG: [[BPC2:%.+]] = bitcast i8** [[BP2]] to %class.S**
// CK34-DAG: [[PC2:%.+]] = bitcast i8** [[P2]] to %class.C** // CK34-DAG: [[PC2:%.+]] = bitcast i8** [[P2]] to %class.C**
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC2]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC2]],
// CK34-DAG: store %class.C* [[C_END:%.+]], %class.C** [[PC2]], // CK34-DAG: store %class.C* [[C_END:%.+]], %class.C** [[PC2]],
// CK34-DAG: store i64 [[B_SIZE:%.+]], i64* [[S2]], // CK34-DAG: store i64 [[B_SIZE:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE_TO]], i32 0, i32 2),
// CK34-DAG: store i8* null, i8** [[MF2]], // CK34-DAG: store i8* null, i8** [[MF2]],
// CK34-DAG: [[C_END]] = getelementptr %class.C, %class.C* [[C_ADDR]], i{{.+}} 1 // CK34-DAG: [[C_END]] = getelementptr %class.C, %class.C* [[C_ADDR]], i{{.+}} 1
// CK34-DAG: [[B_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK34-DAG: [[B_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK34-DAG: [[SZ]] = sub i64 [[S_END_INTPTR:%.+]], [[C_END_INTPTR:%.+]] // CK34-DAG: [[SZ]] = sub i64 [[S_END_INTPTR:%.+]], [[C_END_INTPTR:%.+]]
// CK34-DAG: [[C_END_INTPTR]] = ptrtoint i8* [[C_END_VOID:%.+]] to i64 // CK34-DAG: [[C_END_INTPTR]] = ptrtoint i8* [[C_END_VOID:%.+]] to i64
// CK34-DAG: [[S_END_INTPTR]] = ptrtoint i8* [[S_END_VOID:%.+]] to i64 // CK34-DAG: [[S_END_INTPTR]] = ptrtoint i8* [[S_END_VOID:%.+]] to i64
// CK34-DAG: [[C_END_VOID]] = bitcast %class.C* [[C_END]] to i8* // CK34-DAG: [[C_END_VOID]] = bitcast %class.C* [[C_END]] to i8*
// CK34-DAG: [[S_END_VOID]] = getelementptr i8, i8* [[S_LAST:%.+]], i{{.+}} 1 // CK34-DAG: [[S_END_VOID]] = getelementptr i8, i8* [[S_LAST:%.+]], i{{.+}} 1
// CK34-64-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOID:%.+]], i64 31 // CK34-64-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOID:%.+]], i64 31
// CK34-32-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOID:%.+]], i32 15 // CK34-32-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOID:%.+]], i32 15
// CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8* // CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8*
// pass MEMBER_OF_1 | TO | IMPLICIT | MAPPER {&s, &s.c, 16} to copy the data of s.c. // pass MEMBER_OF_1 | TO | IMPLICIT | MAPPER {&s, &s.c, 16} to copy the data of s.c.
// CK34-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3 // CK34-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3
// CK34-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3 // CK34-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3
// CK34-DAG: [[S3:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 3
// CK34-DAG: [[MF3:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 3 // CK34-DAG: [[MF3:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 3
// CK34-DAG: [[BPC3:%.+]] = bitcast i8** [[BP3]] to %class.S** // CK34-DAG: [[BPC3:%.+]] = bitcast i8** [[BP3]] to %class.S**
// CK34-DAG: [[PC3:%.+]] = bitcast i8** [[P3]] to %class.C** // CK34-DAG: [[PC3:%.+]] = bitcast i8** [[P3]] to %class.C**
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC3]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC3]],
// CK34-DAG: store %class.C* [[C_ADDR:%.+]], %class.C** [[PC3]], // CK34-DAG: store %class.C* [[C_ADDR:%.+]], %class.C** [[PC3]],
// CK34-64-DAG: store i64 16, i64* [[S3]],
// CK34-32-DAG: store i64 8, i64* [[S3]],
// CK34-DAG: store i8* bitcast (void (i8*, i8*, i8*, i64, i64, i8*)* [[C_DEFAULT_MAPPER:@.+]] to i8*), i8** [[MF3]], // CK34-DAG: store i8* bitcast (void (i8*, i8*, i8*, i64, i64, i8*)* [[C_DEFAULT_MAPPER:@.+]] to i8*), i8** [[MF3]],
// CK34-64-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 2 // CK34-64-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 2
// CK34-32-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1 // CK34-32-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1
#pragma omp target map(to: s) #pragma omp target map(to: s)
s.foo(); s.foo();
// CK34 : call void // CK34 : call void
// CK34-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{.+}}, i32 4, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[MTYPE_FROM]]{{.+}}, i8** null, i8** [[GEPMF:%.+]]) // CK34-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{.+}}, i32 4, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[SIZE_FROM]]{{.+}}, {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[MTYPE_FROM]]{{.+}}, i8** null, i8** [[GEPMF:%.+]])
// CK34-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK34-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK34-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK34-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK34-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK34-DAG: [[GEPMF]] = bitcast [4 x i8*]* [[MF:%.+]] to i8** // CK34-DAG: [[GEPMF]] = bitcast [4 x i8*]* [[MF:%.+]] to i8**
// pass TARGET_PARAM {&s, &s, ((void*)(&s+1)-(void*)&s)} // pass TARGET_PARAM {&s, &s, ((void*)(&s+1)-(void*)&s)}
// CK34-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK34-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK34-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK34-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK34-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK34-DAG: [[MF0:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 0 // CK34-DAG: [[MF0:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 0
// CK34-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to %class.S** // CK34-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to %class.S**
// CK34-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to %class.S** // CK34-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to %class.S**
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC0]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC0]],
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC0]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC0]],
// CK34-DAG: store i64 [[S_SIZE:%.+]], i64* [[S0]], // CK34-DAG: store i64 [[S_SIZE:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE_FROM]], i32 0, i32 0),
// CK34-DAG: store i8* null, i8** [[MF0]], // CK34-DAG: store i8* null, i8** [[MF0]],
// CK34-DAG: [[S_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK34-DAG: [[S_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK34-DAG: [[SZ]] = sub i64 [[S_1_INTPTR:%.+]], [[S_INTPTR:%.+]] // CK34-DAG: [[SZ]] = sub i64 [[S_1_INTPTR:%.+]], [[S_INTPTR:%.+]]
// CK34-DAG: [[S_1_INTPTR]] = ptrtoint i8* [[S_1_VOID:%.+]] to i64 // CK34-DAG: [[S_1_INTPTR]] = ptrtoint i8* [[S_1_VOID:%.+]] to i64
// CK34-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64 // CK34-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64
// CK34-DAG: [[S_1_VOID]] = bitcast %class.S* [[S_1:%.+]] to i8* // CK34-DAG: [[S_1_VOID]] = bitcast %class.S* [[S_1:%.+]] to i8*
// CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8* // CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8*
// CK34-DAG: [[S_1]] = getelementptr %class.S, %class.S* [[S_ADDR]], i32 1 // CK34-DAG: [[S_1]] = getelementptr %class.S, %class.S* [[S_ADDR]], i32 1
// pass MEMBER_OF_1 | FROM {&s, &s, ((void*)(&s.a+1)-(void*)&s)} to copy the data of s.a. // pass MEMBER_OF_1 | FROM {&s, &s, ((void*)(&s.a+1)-(void*)&s)} to copy the data of s.a.
// CK34-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK34-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK34-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK34-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK34-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK34-DAG: [[MF1:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 1 // CK34-DAG: [[MF1:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 1
// CK34-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to %class.S** // CK34-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to %class.S**
// CK34-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to %class.S** // CK34-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to %class.S**
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC1]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC1]],
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC1]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC1]],
// CK34-DAG: store i64 [[A_SIZE:%.+]], i64* [[S1]], // CK34-DAG: store i64 [[A_SIZE:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE_FROM]], i32 0, i32 1),
// CK34-DAG: store i8* null, i8** [[MF1]], // CK34-DAG: store i8* null, i8** [[MF1]],
// CK34-DAG: [[A_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK34-DAG: [[A_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK34-DAG: [[SZ]] = sub i64 [[C_BEGIN_INTPTR:%.+]], [[S_INTPTR:%.+]] // CK34-DAG: [[SZ]] = sub i64 [[C_BEGIN_INTPTR:%.+]], [[S_INTPTR:%.+]]
// CK34-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64 // CK34-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64
// CK34-DAG: [[C_BEGIN_INTPTR]] = ptrtoint i8* [[C_BEGIN_VOID:%.+]] to i64 // CK34-DAG: [[C_BEGIN_INTPTR]] = ptrtoint i8* [[C_BEGIN_VOID:%.+]] to i64
// CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8* // CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8*
// CK34-DAG: [[C_BEGIN_VOID]] = bitcast %class.C* [[C_ADDR:%.+]] to i8* // CK34-DAG: [[C_BEGIN_VOID]] = bitcast %class.C* [[C_ADDR:%.+]] to i8*
// CK34-64-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 2 // CK34-64-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 2
// CK34-32-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1 // CK34-32-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1
// pass MEMBER_OF_1 | FROM {&s, &s.c+1, ((void*)(&s)+31+1-(void*)(&s.c+1))} to copy the data of s.b. // pass MEMBER_OF_1 | FROM {&s, &s.c+1, ((void*)(&s)+31+1-(void*)(&s.c+1))} to copy the data of s.b.
// CK34-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK34-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK34-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK34-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK34-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK34-DAG: [[MF2:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 2 // CK34-DAG: [[MF2:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 2
// CK34-DAG: [[BPC2:%.+]] = bitcast i8** [[BP2]] to %class.S** // CK34-DAG: [[BPC2:%.+]] = bitcast i8** [[BP2]] to %class.S**
// CK34-DAG: [[PC2:%.+]] = bitcast i8** [[P2]] to %class.C** // CK34-DAG: [[PC2:%.+]] = bitcast i8** [[P2]] to %class.C**
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC2]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC2]],
// CK34-DAG: store %class.C* [[C_END:%.+]], %class.C** [[PC2]], // CK34-DAG: store %class.C* [[C_END:%.+]], %class.C** [[PC2]],
// CK34-DAG: store i64 [[B_SIZE:%.+]], i64* [[S2]], // CK34-DAG: store i64 [[B_SIZE:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE_FROM]], i32 0, i32 2),
// CK34-DAG: store i8* null, i8** [[MF2]], // CK34-DAG: store i8* null, i8** [[MF2]],
// CK34-DAG: [[C_END]] = getelementptr %class.C, %class.C* [[C_ADDR]], i{{.+}} 1 // CK34-DAG: [[C_END]] = getelementptr %class.C, %class.C* [[C_ADDR]], i{{.+}} 1
// CK34-DAG: [[B_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK34-DAG: [[B_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK34-DAG: [[SZ]] = sub i64 [[S_END_INTPTR:%.+]], [[C_END_INTPTR:%.+]] // CK34-DAG: [[SZ]] = sub i64 [[S_END_INTPTR:%.+]], [[C_END_INTPTR:%.+]]
// CK34-DAG: [[C_END_INTPTR]] = ptrtoint i8* [[C_END_VOID:%.+]] to i64 // CK34-DAG: [[C_END_INTPTR]] = ptrtoint i8* [[C_END_VOID:%.+]] to i64
// CK34-DAG: [[S_END_INTPTR]] = ptrtoint i8* [[S_END_VOID:%.+]] to i64 // CK34-DAG: [[S_END_INTPTR]] = ptrtoint i8* [[S_END_VOID:%.+]] to i64
// CK34-DAG: [[C_END_VOID]] = bitcast %class.C* [[C_END]] to i8* // CK34-DAG: [[C_END_VOID]] = bitcast %class.C* [[C_END]] to i8*
// CK34-DAG: [[S_END_VOID]] = getelementptr i8, i8* [[S_LAST:%.+]], i{{.+}} 1 // CK34-DAG: [[S_END_VOID]] = getelementptr i8, i8* [[S_LAST:%.+]], i{{.+}} 1
// CK34-64-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOID:%.+]], i64 31 // CK34-64-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOID:%.+]], i64 31
// CK34-32-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOID:%.+]], i32 15 // CK34-32-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOID:%.+]], i32 15
// CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8* // CK34-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8*
// pass MEMBER_OF_1 | FROM | IMPLICIT | MAPPER {&s, &s.c, 16} to copy the data of s.c. // pass MEMBER_OF_1 | FROM | IMPLICIT | MAPPER {&s, &s.c, 16} to copy the data of s.c.
// CK34-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3 // CK34-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3
// CK34-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3 // CK34-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3
// CK34-DAG: [[S3:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 3
// CK34-DAG: [[MF3:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 3 // CK34-DAG: [[MF3:%.+]] = getelementptr inbounds {{.+}}[[MF]], i{{.+}} 0, i{{.+}} 3
// CK34-DAG: [[BPC3:%.+]] = bitcast i8** [[BP3]] to %class.S** // CK34-DAG: [[BPC3:%.+]] = bitcast i8** [[BP3]] to %class.S**
// CK34-DAG: [[PC3:%.+]] = bitcast i8** [[P3]] to %class.C** // CK34-DAG: [[PC3:%.+]] = bitcast i8** [[P3]] to %class.C**
// CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC3]], // CK34-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC3]],
// CK34-DAG: store %class.C* [[C_ADDR:%.+]], %class.C** [[PC3]], // CK34-DAG: store %class.C* [[C_ADDR:%.+]], %class.C** [[PC3]],
// CK34-64-DAG: store i64 16, i64* [[S3]],
// CK34-32-DAG: store i64 8, i64* [[S3]],
// CK34-DAG: store i8* bitcast (void (i8*, i8*, i8*, i64, i64, i8*)* [[C_DEFAULT_MAPPER]] to i8*), i8** [[MF3]], // CK34-DAG: store i8* bitcast (void (i8*, i8*, i8*, i64, i64, i8*)* [[C_DEFAULT_MAPPER]] to i8*), i8** [[MF3]],
// CK34-64-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 2 // CK34-64-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 2
// CK34-32-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1 // CK34-32-DAG: [[C_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1
#pragma omp target map(from: s) #pragma omp target map(from: s)
s.foo(); s.foo();
} }
#endif // CK34 #endif // CK34
#endif #endif

clang/test/OpenMP/target_map_codegen_35.cpp

Show All 21 Lines
class S { class S {
public: public:
S(double &b) : b(b) {} S(double &b) : b(b) {}
int a; int a;
double &b; double &b;
void foo(); void foo();
}; };
// CK35-DAG: [[SIZE_TO:@.+]] = private {{.*}}global [4 x i64] [i64 0, i64 0, i64 0, i64 8]
// TARGET_PARAM = 0x20 // TARGET_PARAM = 0x20
// MEMBER_OF_1 | TO = 0x1000000000001 // MEMBER_OF_1 | TO = 0x1000000000001
// MEMBER_OF_1 | PTR_AND_OBJ | TO = 0x1000000000011 // MEMBER_OF_1 | PTR_AND_OBJ | TO = 0x1000000000011
// CK35-DAG: [[MTYPE_TO:@.+]] = {{.+}}constant [4 x i64] [i64 [[#0x20]], i64 [[#0x1000000000001]], i64 [[#0x1000000000001]], i64 [[#0x1000000000011]]] // CK35-DAG: [[MTYPE_TO:@.+]] = {{.+}}constant [4 x i64] [i64 [[#0x20]], i64 [[#0x1000000000001]], i64 [[#0x1000000000001]], i64 [[#0x1000000000011]]]
// CK35-DAG: [[SIZE_FROM:@.+]] = private {{.*}}global [2 x i64] [i64 0, i64 8]
// TARGET_PARAM = 0x20 // TARGET_PARAM = 0x20
// MEMBER_OF_1 | PTR_AND_OBJ | FROM = 0x1000000000012 // MEMBER_OF_1 | PTR_AND_OBJ | FROM = 0x1000000000012
// CK35-DAG: [[MTYPE_FROM:@.+]] = {{.+}}constant [2 x i64] [i64 [[#0x20]], i64 [[#0x1000000000012]]] // CK35-DAG: [[MTYPE_FROM:@.+]] = {{.+}}constant [2 x i64] [i64 [[#0x20]], i64 [[#0x1000000000012]]]
void ref_map() { void ref_map() {
double b; double b;
S s(b); S s(b);
// CK35-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{.+}}, i32 4, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[MTYPE_TO]]{{.+}}, i8** null, i8** null) // CK35-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{.+}}, i32 4, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[SIZE_TO]]{{.+}}, {{.+}}getelementptr {{.+}}[4 x i{{.+}}]* [[MTYPE_TO]]{{.+}}, i8** null, i8** null)
// CK35-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK35-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK35-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK35-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK35-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// pass TARGET_PARAM {&s, &s, ((void*)(&s+1)-(void*)&s)} // pass TARGET_PARAM {&s, &s, ((void*)(&s+1)-(void*)&s)}
// CK35-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK35-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK35-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK35-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK35-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK35-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to %class.S** // CK35-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to %class.S**
// CK35-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to %class.S** // CK35-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to %class.S**
// CK35-DAG: store %class.S* [[S_ADDR:%.+]], %class.S** [[BPC0]], // CK35-DAG: store %class.S* [[S_ADDR:%.+]], %class.S** [[BPC0]],
// CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC0]], // CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC0]],
// CK35-DAG: store i64 [[S_SIZE:%.+]], i64* [[S0]], // CK35-DAG: store i64 [[S_SIZE:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE_TO]], i32 0, i32 0),
// CK35-DAG: [[S_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK35-DAG: [[S_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK35-DAG: [[SZ]] = sub i64 [[S_1_INTPTR:%.+]], [[S_INTPTR:%.+]] // CK35-DAG: [[SZ]] = sub i64 [[S_1_INTPTR:%.+]], [[S_INTPTR:%.+]]
// CK35-DAG: [[S_1_INTPTR]] = ptrtoint i8* [[S_1_VOID:%.+]] to i64 // CK35-DAG: [[S_1_INTPTR]] = ptrtoint i8* [[S_1_VOID:%.+]] to i64
// CK35-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64 // CK35-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64
// CK35-DAG: [[S_1_VOID]] = bitcast %class.S* [[S_1:%.+]] to i8* // CK35-DAG: [[S_1_VOID]] = bitcast %class.S* [[S_1:%.+]] to i8*
// CK35-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8* // CK35-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8*
// CK35-DAG: [[S_1]] = getelementptr %class.S, %class.S* [[S_ADDR]], i32 1 // CK35-DAG: [[S_1]] = getelementptr %class.S, %class.S* [[S_ADDR]], i32 1
// pass MEMBER_OF_1 | TO {&s, &s, ((void*)(&s.a+1)-(void*)&s)} to copy the data of s.a. // pass MEMBER_OF_1 | TO {&s, &s, ((void*)(&s.a+1)-(void*)&s)} to copy the data of s.a.
// CK35-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK35-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK35-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK35-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK35-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK35-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to %class.S** // CK35-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to %class.S**
// CK35-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to %class.S** // CK35-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to %class.S**
// CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC1]], // CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC1]],
// CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC1]], // CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[PC1]],
// CK35-DAG: store i64 [[A_SIZE:%.+]], i64* [[S1]], // CK35-DAG: store i64 [[A_SIZE:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE_TO]], i32 0, i32 1),
// CK35-DAG: [[A_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK35-DAG: [[A_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK35-DAG: [[SZ]] = sub i64 [[B_BEGIN_INTPTR:%.+]], [[S_INTPTR:%.+]] // CK35-DAG: [[SZ]] = sub i64 [[B_BEGIN_INTPTR:%.+]], [[S_INTPTR:%.+]]
// CK35-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64 // CK35-DAG: [[S_INTPTR]] = ptrtoint i8* [[S_VOID:%.+]] to i64
// CK35-DAG: [[B_BEGIN_INTPTR]] = ptrtoint i8* [[B_BEGIN_VOID:%.+]] to i64 // CK35-DAG: [[B_BEGIN_INTPTR]] = ptrtoint i8* [[B_BEGIN_VOID:%.+]] to i64
// CK35-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8* // CK35-DAG: [[S_VOID]] = bitcast %class.S* [[S_ADDR]] to i8*
// CK35-DAG: [[B_BEGIN_VOID]] = bitcast double** [[B_ADDR:%.+]] to i8* // CK35-DAG: [[B_BEGIN_VOID]] = bitcast double** [[B_ADDR:%.+]] to i8*
// CK35-DAG: [[B_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1 // CK35-DAG: [[B_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1
// pass MEMBER_OF_1 | TO {&s, &s.b+1, ((void*)(&s+1)-(void*)(&s.b+1))} to copy the data of remainder of s. // pass MEMBER_OF_1 | TO {&s, &s.b+1, ((void*)(&s+1)-(void*)(&s.b+1))} to copy the data of remainder of s.
// CK35-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK35-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK35-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK35-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK35-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 2
// CK35-DAG: [[BPC2:%.+]] = bitcast i8** [[BP2]] to %class.S** // CK35-DAG: [[BPC2:%.+]] = bitcast i8** [[BP2]] to %class.S**
// CK35-DAG: [[PC2:%.+]] = bitcast i8** [[P2]] to double*** // CK35-DAG: [[PC2:%.+]] = bitcast i8** [[P2]] to double***
// CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC2]], // CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC2]],
// CK35-DAG: store double** [[B_END:%.+]], double*** [[PC2]], // CK35-DAG: store double** [[B_END:%.+]], double*** [[PC2]],
// CK35-DAG: store i64 [[REM_SIZE:%.+]], i64* [[S2]], // CK35-DAG: store i64 [[REM_SIZE:%.+]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* [[SIZE_TO]], i32 0, i32 2),
// CK35-DAG: [[B_END]] = getelementptr double*, double** [[B_ADDR]], i{{.+}} 1 // CK35-DAG: [[B_END]] = getelementptr double*, double** [[B_ADDR]], i{{.+}} 1
// CK35-DAG: [[REM_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK35-DAG: [[REM_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK35-DAG: [[SZ]] = sub i64 [[S_END_INTPTR:%.+]], [[B_END_INTPTR:%.+]] // CK35-DAG: [[SZ]] = sub i64 [[S_END_INTPTR:%.+]], [[B_END_INTPTR:%.+]]
// CK35-DAG: [[B_END_INTPTR]] = ptrtoint i8* [[B_END_VOID:%.+]] to i64 // CK35-DAG: [[B_END_INTPTR]] = ptrtoint i8* [[B_END_VOID:%.+]] to i64
// CK35-DAG: [[S_END_INTPTR]] = ptrtoint i8* [[S_END_VOID:%.+]] to i64 // CK35-DAG: [[S_END_INTPTR]] = ptrtoint i8* [[S_END_VOID:%.+]] to i64
// CK35-DAG: [[B_END_VOID]] = bitcast double** [[B_END]] to i8* // CK35-DAG: [[B_END_VOID]] = bitcast double** [[B_END]] to i8*
// CK35-DAG: [[S_END_VOID]] = getelementptr i8, i8* [[S_LAST:%.+]], i{{.+}} 1 // CK35-DAG: [[S_END_VOID]] = getelementptr i8, i8* [[S_LAST:%.+]], i{{.+}} 1
// CK35-64-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOIDPTR:%.+]], i64 15 // CK35-64-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOIDPTR:%.+]], i64 15
// CK35-32-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOIDPTR:%.+]], i32 7 // CK35-32-DAG: [[S_LAST]] = getelementptr i8, i8* [[S_VOIDPTR:%.+]], i32 7
// CK35-DAG: [[S_VOIDPTR]] = bitcast %class.S* [[S_ADDR]] to i8* // CK35-DAG: [[S_VOIDPTR]] = bitcast %class.S* [[S_ADDR]] to i8*
// pass MEMBER_OF_1 | PTR_AND_OBJ | TO {&s, &s.b, 8|4} to copy the data of s.b. // pass MEMBER_OF_1 | PTR_AND_OBJ | TO {&s, &s.b, 8|4} to copy the data of s.b.
// CK35-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3 // CK35-DAG: [[BP3:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 3
// CK35-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3 // CK35-DAG: [[P3:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 3
// CK35-DAG: [[S3:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 3
// CK35-DAG: [[BPC3:%.+]] = bitcast i8** [[BP3]] to %class.S** // CK35-DAG: [[BPC3:%.+]] = bitcast i8** [[BP3]] to %class.S**
// CK35-DAG: [[PC3:%.+]] = bitcast i8** [[P3]] to double** // CK35-DAG: [[PC3:%.+]] = bitcast i8** [[P3]] to double**
// CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC3]], // CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC3]],
// CK35-DAG: store double* [[B_ADDR:%.+]], double** [[PC3]], // CK35-DAG: store double* [[B_ADDR:%.+]], double** [[PC3]],
// CK35-DAG: store i64 8, i64* [[S3]],
// CK35-DAG: [[B_ADDR]] = load double*, double** [[B_REF:%.+]], // CK35-DAG: [[B_ADDR]] = load double*, double** [[B_REF:%.+]],
// CK35-DAG: [[B_REF]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1 // CK35-DAG: [[B_REF]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1
#pragma omp target map(to: s, s.b) #pragma omp target map(to: s, s.b)
s.foo(); s.foo();
// CK35 : call void // CK35 : call void
// CK35-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{.+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE_FROM]]{{.+}}, i8** null, i8** null) // CK35-DAG: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{.+}}, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE_FROM]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE_FROM]]{{.+}}, i8** null, i8** null)
// CK35-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK35-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK35-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK35-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK35-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// pass TARGET_PARAM {&s, &s.b, ((void*)(&s.b+1)-(void*)&s.b)} // pass TARGET_PARAM {&s, &s.b, ((void*)(&s.b+1)-(void*)&s.b)}
// CK35-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK35-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK35-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK35-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK35-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK35-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to %class.S** // CK35-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to %class.S**
// CK35-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double*** // CK35-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double***
// CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC0]], // CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC0]],
// CK35-DAG: store double** [[SB_ADDR:%.+]], double*** [[PC0]], // CK35-DAG: store double** [[SB_ADDR:%.+]], double*** [[PC0]],
// CK35-DAG: store i64 [[B_SIZE:%.+]], i64* [[S0]], // CK35-DAG: store i64 [[B_SIZE:%.+]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE_FROM]], i32 0, i32 0),
// CK35-DAG: [[B_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK35-DAG: [[B_SIZE]] = sdiv exact i64 [[SZ:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK35-DAG: [[SZ]] = sub i64 [[SB_1_INTPTR:%.+]], [[SB_INTPTR:%.+]] // CK35-DAG: [[SZ]] = sub i64 [[SB_1_INTPTR:%.+]], [[SB_INTPTR:%.+]]
// CK35-DAG: [[SB_1_INTPTR]] = ptrtoint i8* [[SB_1_VOID:%.+]] to i64 // CK35-DAG: [[SB_1_INTPTR]] = ptrtoint i8* [[SB_1_VOID:%.+]] to i64
// CK35-DAG: [[SB_INTPTR]] = ptrtoint i8* [[SB_VOID:%.+]] to i64 // CK35-DAG: [[SB_INTPTR]] = ptrtoint i8* [[SB_VOID:%.+]] to i64
// CK35-DAG: [[SB_1_VOID]] = bitcast double** [[SB_1:%.+]] to i8* // CK35-DAG: [[SB_1_VOID]] = bitcast double** [[SB_1:%.+]] to i8*
// CK35-DAG: [[SB_VOID]] = bitcast double** [[SB_ADDR:%.+]] to i8* // CK35-DAG: [[SB_VOID]] = bitcast double** [[SB_ADDR:%.+]] to i8*
// CK35-DAG: [[SB_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1 // CK35-DAG: [[SB_ADDR]] = getelementptr inbounds %class.S, %class.S* [[S_ADDR]], i32 0, i32 1
// CK35-DAG: [[SB_1]] = getelementptr double*, double** [[SB_ADDR]], i{{.+}} 1 // CK35-DAG: [[SB_1]] = getelementptr double*, double** [[SB_ADDR]], i{{.+}} 1
// pass MEMBER_OF_1 | PTR_AND_OBJ | FROM {&s, &s.b, 8|4} to copy the data of s.c. // pass MEMBER_OF_1 | PTR_AND_OBJ | FROM {&s, &s.b, 8|4} to copy the data of s.c.
// CK35-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK35-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK35-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK35-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK35-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 1
// CK35-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to %class.S** // CK35-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to %class.S**
// CK35-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to double** // CK35-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to double**
// CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC1]], // CK35-DAG: store %class.S* [[S_ADDR]], %class.S** [[BPC1]],
// CK35-DAG: store double* [[B_ADDR:%.+]], double** [[PC1]], // CK35-DAG: store double* [[B_ADDR:%.+]], double** [[PC1]],
// CK35-DAG: store i64 8, i64* [[S1]],
// CK35-DAG: [[B_ADDR]] = load double*, double** [[SB_ADDR]], // CK35-DAG: [[B_ADDR]] = load double*, double** [[SB_ADDR]],
#pragma omp target map(from: s.b) #pragma omp target map(from: s.b)
s.foo(); s.foo();
} }
#endif // CK35 #endif // CK35
#endif #endif

clang/test/OpenMP/target_map_codegen_hold.cpp

// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --check-globals --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" --global-value-regex ".offload_maptypes.*" ".offload_sizes.*" --global-hex-value-regex ".offload_maptypes.*" // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --check-globals --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" --prefix-filecheck-ir-name _ --global-value-regex ".offload_maptypes.*" ".offload_sizes.*" --global-hex-value-regex ".offload_maptypes.*"
// expected-no-diagnostics // expected-no-diagnostics
#ifndef HEADER #ifndef HEADER
#define HEADER #define HEADER
//-------------------------------------------------- //--------------------------------------------------
// With -DUSE. // With -DUSE.
//-------------------------------------------------- //--------------------------------------------------
▲ Show 20 Lines • Show All 68 Lines • ▼ Show 20 Lines
// ALWAYS = 0x4 // ALWAYS = 0x4
// TARGET_PARAM = 0x20 // TARGET_PARAM = 0x20
// CLOSE = 0x400 // CLOSE = 0x400
// OMPX_HOLD = 0x2000 // OMPX_HOLD = 0x2000
// MEMBER_OF_1 = 0x1000000000000 // MEMBER_OF_1 = 0x1000000000000
// MEMBER_OF_5 = 0x5000000000000 // MEMBER_OF_5 = 0x5000000000000
//. //.
// CHECK-USE-PPC64LE: @.offload_sizes = private unnamed_addr global [7 x i64] [i64 0, i64 4, i64 4, i64 4, i64 0, i64 4, i64 4]
// CHECK-USE-PPC64LE: @.offload_maptypes = private unnamed_addr constant [7 x i64] [i64 [[#0x2020]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]], i64 [[#0x2023]], i64 [[#0x2020]], i64 [[#0x5000000002003]], i64 [[#0x5000000002003]]] // CHECK-USE-PPC64LE: @.offload_maptypes = private unnamed_addr constant [7 x i64] [i64 [[#0x2020]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]], i64 [[#0x2023]], i64 [[#0x2020]], i64 [[#0x5000000002003]], i64 [[#0x5000000002003]]]
// CHECK-USE-PPC64LE: @.offload_sizes = private unnamed_addr constant [1 x i64] [i64 4] // CHECK-USE-PPC64LE: @.offload_sizes.1 = private unnamed_addr constant [1 x i64] [i64 4]
// CHECK-USE-PPC64LE: @.offload_maptypes.1 = private unnamed_addr constant [1 x i64] [i64 [[#0x2427]]] // CHECK-USE-PPC64LE: @.offload_maptypes.2 = private unnamed_addr constant [1 x i64] [i64 [[#0x2427]]]
// CHECK-USE-PPC64LE: @.offload_maptypes.2 = private unnamed_addr constant [3 x i64] [i64 [[#0x2020]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]]] // CHECK-USE-PPC64LE: @.offload_sizes.3 = private unnamed_addr global [3 x i64] [i64 0, i64 4, i64 4]
// CHECK-USE-PPC64LE: @.offload_maptypes.4 = private unnamed_addr constant [3 x i64] [i64 [[#0x2020]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]]]
//. //.
// CHECK-USE-I386: @.offload_sizes = private unnamed_addr global [7 x i64] [i64 0, i64 4, i64 4, i64 4, i64 0, i64 4, i64 4]
// CHECK-USE-I386: @.offload_maptypes = private unnamed_addr constant [7 x i64] [i64 [[#0x2020]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]], i64 [[#0x2023]], i64 [[#0x2020]], i64 [[#0x5000000002003]], i64 [[#0x5000000002003]]] // CHECK-USE-I386: @.offload_maptypes = private unnamed_addr constant [7 x i64] [i64 [[#0x2020]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]], i64 [[#0x2023]], i64 [[#0x2020]], i64 [[#0x5000000002003]], i64 [[#0x5000000002003]]]
// CHECK-USE-I386: @.offload_sizes = private unnamed_addr constant [1 x i64] [i64 4] // CHECK-USE-I386: @.offload_sizes.1 = private unnamed_addr constant [1 x i64] [i64 4]
// CHECK-USE-I386: @.offload_maptypes.1 = private unnamed_addr constant [1 x i64] [i64 [[#0x2427]]] // CHECK-USE-I386: @.offload_maptypes.2 = private unnamed_addr constant [1 x i64] [i64 [[#0x2427]]]
// CHECK-USE-I386: @.offload_maptypes.2 = private unnamed_addr constant [3 x i64] [i64 [[#0x2020]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]]] // CHECK-USE-I386: @.offload_sizes.3 = private unnamed_addr global [3 x i64] [i64 0, i64 4, i64 4]
// CHECK-USE-I386: @.offload_maptypes.4 = private unnamed_addr constant [3 x i64] [i64 [[#0x2020]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]]]
//. //.
// CHECK-NOUSE-PPC64LE: @.offload_sizes = private unnamed_addr global [7 x i64] [i64 0, i64 4, i64 4, i64 4, i64 0, i64 4, i64 4]
// CHECK-NOUSE-PPC64LE: @.offload_maptypes = private unnamed_addr constant [7 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]], i64 [[#0x2003]], i64 [[#0x2000]], i64 [[#0x5000000002003]], i64 [[#0x5000000002003]]] // CHECK-NOUSE-PPC64LE: @.offload_maptypes = private unnamed_addr constant [7 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]], i64 [[#0x2003]], i64 [[#0x2000]], i64 [[#0x5000000002003]], i64 [[#0x5000000002003]]]
// CHECK-NOUSE-PPC64LE: @.offload_sizes = private unnamed_addr constant [1 x i64] [i64 4] // CHECK-NOUSE-PPC64LE: @.offload_sizes.1 = private unnamed_addr constant [1 x i64] [i64 4]
// CHECK-NOUSE-PPC64LE: @.offload_maptypes.1 = private unnamed_addr constant [1 x i64] [i64 [[#0x2407]]] // CHECK-NOUSE-PPC64LE: @.offload_maptypes.2 = private unnamed_addr constant [1 x i64] [i64 [[#0x2407]]]
// CHECK-NOUSE-PPC64LE: @.offload_maptypes.2 = private unnamed_addr constant [3 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]]] // CHECK-NOUSE-PPC64LE: @.offload_sizes.3 = private unnamed_addr global [3 x i64] [i64 0, i64 4, i64 4]
// CHECK-NOUSE-PPC64LE: @.offload_maptypes.4 = private unnamed_addr constant [3 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]]]
//. //.
// CHECK-NOUSE-I386: @.offload_sizes = private unnamed_addr global [7 x i64] [i64 0, i64 4, i64 4, i64 4, i64 0, i64 4, i64 4]
// CHECK-NOUSE-I386: @.offload_maptypes = private unnamed_addr constant [7 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]], i64 [[#0x2003]], i64 [[#0x2000]], i64 [[#0x5000000002003]], i64 [[#0x5000000002003]]] // CHECK-NOUSE-I386: @.offload_maptypes = private unnamed_addr constant [7 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]], i64 [[#0x2003]], i64 [[#0x2000]], i64 [[#0x5000000002003]], i64 [[#0x5000000002003]]]
// CHECK-NOUSE-I386: @.offload_sizes = private unnamed_addr constant [1 x i64] [i64 4] // CHECK-NOUSE-I386: @.offload_sizes.1 = private unnamed_addr constant [1 x i64] [i64 4]
// CHECK-NOUSE-I386: @.offload_maptypes.1 = private unnamed_addr constant [1 x i64] [i64 [[#0x2407]]] // CHECK-NOUSE-I386: @.offload_maptypes.2 = private unnamed_addr constant [1 x i64] [i64 [[#0x2407]]]
// CHECK-NOUSE-I386: @.offload_maptypes.2 = private unnamed_addr constant [3 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]]] // CHECK-NOUSE-I386: @.offload_sizes.3 = private unnamed_addr global [3 x i64] [i64 0, i64 4, i64 4]
// CHECK-NOUSE-I386: @.offload_maptypes.4 = private unnamed_addr constant [3 x i64] [i64 [[#0x2000]], i64 [[#0x1000000002003]], i64 [[#0x1000000002003]]]
//. //.
struct ST { struct ST {
int i; int i;
int j; int j;
void test_present_members(); void test_present_members();
}; };
// CHECK-USE-PPC64LE-LABEL: @_Z20explicit_maps_singlei( // CHECK-USE-PPC64LE-LABEL: @_Z20explicit_maps_singlei(
// CHECK-USE-PPC64LE-NEXT: entry: // CHECK-USE-PPC64LE-NEXT: entry:
// CHECK-USE-PPC64LE-NEXT: [[II_ADDR:%.*]] = alloca i32, align 4 // CHECK-USE-PPC64LE-NEXT: [[II_ADDR:%.*]] = alloca i32, align 4
// CHECK-USE-PPC64LE-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK-USE-PPC64LE-NEXT: [[A:%.*]] = alloca i32, align 4
// CHECK-USE-PPC64LE-NEXT: [[ST1:%.*]] = alloca [[STRUCT_ST:%.*]], align 4 // CHECK-USE-PPC64LE-NEXT: [[ST1:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK-USE-PPC64LE-NEXT: [[ST2:%.*]] = alloca [[STRUCT_ST]], align 4 // CHECK-USE-PPC64LE-NEXT: [[ST2:%.*]] = alloca [[STRUCT_ST]], align 4
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [7 x i8*], align 8 // CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [7 x i8*], align 8
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [7 x i8*], align 8 // CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [7 x i8*], align 8
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [7 x i8*], align 8 // CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [7 x i8*], align 8
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [7 x i64], align 8
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x i8*], align 8 // CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x i8*], align 8
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x i8*], align 8 // CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x i8*], align 8
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x i8*], align 8 // CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x i8*], align 8
// CHECK-USE-PPC64LE-NEXT: store i32 [[II:%.*]], i32* [[II_ADDR]], align 4 // CHECK-USE-PPC64LE-NEXT: store i32 [[II:%.*]], i32* [[II_ADDR]], align 4
// CHECK-USE-PPC64LE-NEXT: [[TMP0:%.*]] = load i32, i32* [[II_ADDR]], align 4 // CHECK-USE-PPC64LE-NEXT: [[TMP0:%.*]] = load i32, i32* [[II_ADDR]], align 4
// CHECK-USE-PPC64LE-NEXT: store i32 [[TMP0]], i32* [[A]], align 4 // CHECK-USE-PPC64LE-NEXT: store i32 [[TMP0]], i32* [[A]], align 4
// CHECK-USE-PPC64LE-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 1 // CHECK-USE-PPC64LE-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 1
Show All 14 Lines
// CHECK-USE-PPC64LE-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]] // CHECK-USE-PPC64LE-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]]
// CHECK-USE-PPC64LE-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK-USE-PPC64LE-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK-USE-PPC64LE-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to %struct.ST** // CHECK-USE-PPC64LE-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to %struct.ST**
// CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP16]], align 8 // CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP16]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** // CHECK-USE-PPC64LE-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32**
// CHECK-USE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP18]], align 8 // CHECK-USE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP18]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 0), align 8
// CHECK-USE-PPC64LE-NEXT: store i64 [[TMP7]], i64* [[TMP19]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK-USE-PPC64LE-NEXT: [[TMP20:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP20]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP20:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK-USE-PPC64LE-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK-USE-PPC64LE-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.ST**
// CHECK-USE-PPC64LE-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.ST** // CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP21]], align 8
// CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP22]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP22:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK-USE-PPC64LE-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK-USE-PPC64LE-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32**
// CHECK-USE-PPC64LE-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32** // CHECK-USE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP23]], align 8
// CHECK-USE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP24]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP24:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK-USE-PPC64LE-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK-USE-PPC64LE-NEXT: store i64 4, i64* [[TMP25]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK-USE-PPC64LE-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK-USE-PPC64LE-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to %struct.ST**
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP26]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK-USE-PPC64LE-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK-USE-PPC64LE-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to %struct.ST** // CHECK-USE-PPC64LE-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32**
// CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP28]], align 8 // CHECK-USE-PPC64LE-NEXT: store i32* [[J]], i32** [[TMP28]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK-USE-PPC64LE-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK-USE-PPC64LE-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32** // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK-USE-PPC64LE-NEXT: store i32* [[J]], i32** [[TMP30]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP30:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK-USE-PPC64LE-NEXT: [[TMP31:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK-USE-PPC64LE-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32**
// CHECK-USE-PPC64LE-NEXT: store i64 4, i64* [[TMP31]], align 8 // CHECK-USE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP31]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK-USE-PPC64LE-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP32]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK-USE-PPC64LE-NEXT: [[TMP33:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK-USE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP33]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32** // CHECK-USE-PPC64LE-NEXT: [[TMP34:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK-USE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP34]], align 8 // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK-USE-PPC64LE-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK-USE-PPC64LE-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK-USE-PPC64LE-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to %struct.ST**
// CHECK-USE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP36]], align 8 // CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP36]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP37:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK-USE-PPC64LE-NEXT: [[TMP37:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK-USE-PPC64LE-NEXT: store i64 4, i64* [[TMP37]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32**
// CHECK-USE-PPC64LE-NEXT: [[TMP38:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK-USE-PPC64LE-NEXT: store i32* [[I1]], i32** [[TMP38]], align 8
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP38]], align 8 // CHECK-USE-PPC64LE-NEXT: store i64 [[TMP14]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP39:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK-USE-PPC64LE-NEXT: [[TMP39:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK-USE-PPC64LE-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to %struct.ST** // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP40]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP40:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK-USE-PPC64LE-NEXT: [[TMP41:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK-USE-PPC64LE-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to %struct.ST**
// CHECK-USE-PPC64LE-NEXT: [[TMP42:%.*]] = bitcast i8** [[TMP41]] to i32** // CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP41]], align 8
// CHECK-USE-PPC64LE-NEXT: store i32* [[I1]], i32** [[TMP42]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP42:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK-USE-PPC64LE-NEXT: [[TMP43:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK-USE-PPC64LE-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK-USE-PPC64LE-NEXT: store i64 [[TMP14]], i64* [[TMP43]], align 8 // CHECK-USE-PPC64LE-NEXT: store i32* [[I1]], i32** [[TMP43]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP44:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK-USE-PPC64LE-NEXT: [[TMP44:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP44]], align 8 // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP44]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP45:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK-USE-PPC64LE-NEXT: [[TMP45:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK-USE-PPC64LE-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to %struct.ST** // CHECK-USE-PPC64LE-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to %struct.ST**
// CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP46]], align 8 // CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP46]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP47:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK-USE-PPC64LE-NEXT: [[TMP47:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK-USE-PPC64LE-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32** // CHECK-USE-PPC64LE-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32**
// CHECK-USE-PPC64LE-NEXT: store i32* [[I1]], i32** [[TMP48]], align 8 // CHECK-USE-PPC64LE-NEXT: store i32* [[J2]], i32** [[TMP48]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP49:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK-USE-PPC64LE-NEXT: [[TMP49:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6
// CHECK-USE-PPC64LE-NEXT: store i64 4, i64* [[TMP49]], align 8 // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP49]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP50:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5 // CHECK-USE-PPC64LE-NEXT: [[TMP50:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP50]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP51:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP51:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK-USE-PPC64LE-NEXT: [[TMP52:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l606.region_id, i32 7, i8** [[TMP50]], i8** [[TMP51]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null)
// CHECK-USE-PPC64LE-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to %struct.ST** // CHECK-USE-PPC64LE-NEXT: [[TMP53:%.*]] = icmp ne i32 [[TMP52]], 0
// CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP52]], align 8 // CHECK-USE-PPC64LE-NEXT: br i1 [[TMP53]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-USE-PPC64LE-NEXT: [[TMP53:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK-USE-PPC64LE-NEXT: [[TMP54:%.*]] = bitcast i8** [[TMP53]] to i32**
// CHECK-USE-PPC64LE-NEXT: store i32* [[J2]], i32** [[TMP54]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP55:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6
// CHECK-USE-PPC64LE-NEXT: store i64 4, i64* [[TMP55]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP56:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP56]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP57:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP58:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP59:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP60:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l654.region_id, i32 7, i8** [[TMP57]], i8** [[TMP58]], i64* [[TMP59]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null)
// CHECK-USE-PPC64LE-NEXT: [[TMP61:%.*]] = icmp ne i32 [[TMP60]], 0
// CHECK-USE-PPC64LE-NEXT: br i1 [[TMP61]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-USE-PPC64LE: omp_offload.failed: // CHECK-USE-PPC64LE: omp_offload.failed:
// CHECK-USE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l654(%struct.ST* [[ST1]], i32* [[A]], %struct.ST* [[ST2]]) #[[ATTR2:[0-9]+]] // CHECK-USE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l606(%struct.ST* [[ST1]], i32* [[A]], %struct.ST* [[ST2]]) #[[ATTR2:[0-9]+]]
// CHECK-USE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-USE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK-USE-PPC64LE: omp_offload.cont: // CHECK-USE-PPC64LE: omp_offload.cont:
// CHECK-USE-PPC64LE-NEXT: [[TMP62:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: [[TMP54:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32** // CHECK-USE-PPC64LE-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to i32**
// CHECK-USE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP63]], align 8 // CHECK-USE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP55]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP64:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: [[TMP56:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32** // CHECK-USE-PPC64LE-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to i32**
// CHECK-USE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP65]], align 8 // CHECK-USE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP57]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP66:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0 // CHECK-USE-PPC64LE-NEXT: [[TMP58:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP66]], align 8 // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP58]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP67:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: [[TMP59:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP68:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: [[TMP60:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP69:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l668.region_id, i32 1, i8** [[TMP67]], i8** [[TMP68]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.1, i32 0, i32 0), i8** null, i8** null) // CHECK-USE-PPC64LE-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l620.region_id, i32 1, i8** [[TMP59]], i8** [[TMP60]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.1, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null)
// CHECK-USE-PPC64LE-NEXT: [[TMP70:%.*]] = icmp ne i32 [[TMP69]], 0 // CHECK-USE-PPC64LE-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0
// CHECK-USE-PPC64LE-NEXT: br i1 [[TMP70]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]] // CHECK-USE-PPC64LE-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
// CHECK-USE-PPC64LE: omp_offload.failed6: // CHECK-USE-PPC64LE: omp_offload.failed6:
// CHECK-USE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l668(i32* [[A]]) #[[ATTR2]] // CHECK-USE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l620(i32* [[A]]) #[[ATTR2]]
// CHECK-USE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT7]] // CHECK-USE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT7]]
// CHECK-USE-PPC64LE: omp_offload.cont7: // CHECK-USE-PPC64LE: omp_offload.cont7:
// CHECK-USE-PPC64LE-NEXT: ret void // CHECK-USE-PPC64LE-NEXT: ret void
// //
// CHECK-USE-I386-LABEL: @_Z20explicit_maps_singlei( // CHECK-USE-I386-LABEL: @_Z20explicit_maps_singlei(
// CHECK-USE-I386-NEXT: entry: // CHECK-USE-I386-NEXT: entry:
// CHECK-USE-I386-NEXT: [[II_ADDR:%.*]] = alloca i32, align 4 // CHECK-USE-I386-NEXT: [[II_ADDR:%.*]] = alloca i32, align 4
// CHECK-USE-I386-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK-USE-I386-NEXT: [[A:%.*]] = alloca i32, align 4
// CHECK-USE-I386-NEXT: [[ST1:%.*]] = alloca [[STRUCT_ST:%.*]], align 4 // CHECK-USE-I386-NEXT: [[ST1:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK-USE-I386-NEXT: [[ST2:%.*]] = alloca [[STRUCT_ST]], align 4 // CHECK-USE-I386-NEXT: [[ST2:%.*]] = alloca [[STRUCT_ST]], align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [7 x i8*], align 4 // CHECK-USE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [7 x i8*], align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [7 x i8*], align 4 // CHECK-USE-I386-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [7 x i8*], align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [7 x i8*], align 4 // CHECK-USE-I386-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [7 x i8*], align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [7 x i64], align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x i8*], align 4 // CHECK-USE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x i8*], align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x i8*], align 4 // CHECK-USE-I386-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x i8*], align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x i8*], align 4 // CHECK-USE-I386-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x i8*], align 4
// CHECK-USE-I386-NEXT: store i32 [[II:%.*]], i32* [[II_ADDR]], align 4 // CHECK-USE-I386-NEXT: store i32 [[II:%.*]], i32* [[II_ADDR]], align 4
// CHECK-USE-I386-NEXT: [[TMP0:%.*]] = load i32, i32* [[II_ADDR]], align 4 // CHECK-USE-I386-NEXT: [[TMP0:%.*]] = load i32, i32* [[II_ADDR]], align 4
// CHECK-USE-I386-NEXT: store i32 [[TMP0]], i32* [[A]], align 4 // CHECK-USE-I386-NEXT: store i32 [[TMP0]], i32* [[A]], align 4
// CHECK-USE-I386-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 1 // CHECK-USE-I386-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 1
Show All 14 Lines
// CHECK-USE-I386-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]] // CHECK-USE-I386-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]]
// CHECK-USE-I386-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK-USE-I386-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK-USE-I386-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to %struct.ST** // CHECK-USE-I386-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to %struct.ST**
// CHECK-USE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP16]], align 4 // CHECK-USE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP16]], align 4
// CHECK-USE-I386-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** // CHECK-USE-I386-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32**
// CHECK-USE-I386-NEXT: store i32* [[I]], i32** [[TMP18]], align 4 // CHECK-USE-I386-NEXT: store i32* [[I]], i32** [[TMP18]], align 4
// CHECK-USE-I386-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-USE-I386-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 0), align 4
// CHECK-USE-I386-NEXT: store i64 [[TMP7]], i64* [[TMP19]], align 4 // CHECK-USE-I386-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP20:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP20]], align 4 // CHECK-USE-I386-NEXT: [[TMP20:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK-USE-I386-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK-USE-I386-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.ST**
// CHECK-USE-I386-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.ST** // CHECK-USE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP21]], align 4
// CHECK-USE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP22]], align 4 // CHECK-USE-I386-NEXT: [[TMP22:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK-USE-I386-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK-USE-I386-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32**
// CHECK-USE-I386-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32** // CHECK-USE-I386-NEXT: store i32* [[I]], i32** [[TMP23]], align 4
// CHECK-USE-I386-NEXT: store i32* [[I]], i32** [[TMP24]], align 4 // CHECK-USE-I386-NEXT: [[TMP24:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK-USE-I386-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK-USE-I386-NEXT: store i64 4, i64* [[TMP25]], align 4 // CHECK-USE-I386-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK-USE-I386-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK-USE-I386-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to %struct.ST**
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK-USE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP26]], align 4
// CHECK-USE-I386-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK-USE-I386-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK-USE-I386-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to %struct.ST** // CHECK-USE-I386-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32**
// CHECK-USE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP28]], align 4 // CHECK-USE-I386-NEXT: store i32* [[J]], i32** [[TMP28]], align 4
// CHECK-USE-I386-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK-USE-I386-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK-USE-I386-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32** // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK-USE-I386-NEXT: store i32* [[J]], i32** [[TMP30]], align 4 // CHECK-USE-I386-NEXT: [[TMP30:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK-USE-I386-NEXT: [[TMP31:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK-USE-I386-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32**
// CHECK-USE-I386-NEXT: store i64 4, i64* [[TMP31]], align 4 // CHECK-USE-I386-NEXT: store i32* [[A]], i32** [[TMP31]], align 4
// CHECK-USE-I386-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK-USE-I386-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP32]], align 4 // CHECK-USE-I386-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK-USE-I386-NEXT: [[TMP33:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK-USE-I386-NEXT: store i32* [[A]], i32** [[TMP33]], align 4
// CHECK-USE-I386-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32** // CHECK-USE-I386-NEXT: [[TMP34:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK-USE-I386-NEXT: store i32* [[A]], i32** [[TMP34]], align 4 // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK-USE-I386-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK-USE-I386-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK-USE-I386-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK-USE-I386-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to %struct.ST**
// CHECK-USE-I386-NEXT: store i32* [[A]], i32** [[TMP36]], align 4 // CHECK-USE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP36]], align 4
// CHECK-USE-I386-NEXT: [[TMP37:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK-USE-I386-NEXT: [[TMP37:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK-USE-I386-NEXT: store i64 4, i64* [[TMP37]], align 4 // CHECK-USE-I386-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32**
// CHECK-USE-I386-NEXT: [[TMP38:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK-USE-I386-NEXT: store i32* [[I1]], i32** [[TMP38]], align 4
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP38]], align 4 // CHECK-USE-I386-NEXT: store i64 [[TMP14]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK-USE-I386-NEXT: [[TMP39:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK-USE-I386-NEXT: [[TMP39:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK-USE-I386-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to %struct.ST** // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK-USE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP40]], align 4 // CHECK-USE-I386-NEXT: [[TMP40:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK-USE-I386-NEXT: [[TMP41:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK-USE-I386-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to %struct.ST**
// CHECK-USE-I386-NEXT: [[TMP42:%.*]] = bitcast i8** [[TMP41]] to i32** // CHECK-USE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP41]], align 4
// CHECK-USE-I386-NEXT: store i32* [[I1]], i32** [[TMP42]], align 4 // CHECK-USE-I386-NEXT: [[TMP42:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK-USE-I386-NEXT: [[TMP43:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK-USE-I386-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK-USE-I386-NEXT: store i64 [[TMP14]], i64* [[TMP43]], align 4 // CHECK-USE-I386-NEXT: store i32* [[I1]], i32** [[TMP43]], align 4
// CHECK-USE-I386-NEXT: [[TMP44:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK-USE-I386-NEXT: [[TMP44:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP44]], align 4 // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP44]], align 4
// CHECK-USE-I386-NEXT: [[TMP45:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK-USE-I386-NEXT: [[TMP45:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK-USE-I386-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to %struct.ST** // CHECK-USE-I386-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to %struct.ST**
// CHECK-USE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP46]], align 4 // CHECK-USE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP46]], align 4
// CHECK-USE-I386-NEXT: [[TMP47:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK-USE-I386-NEXT: [[TMP47:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK-USE-I386-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32** // CHECK-USE-I386-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32**
// CHECK-USE-I386-NEXT: store i32* [[I1]], i32** [[TMP48]], align 4 // CHECK-USE-I386-NEXT: store i32* [[J2]], i32** [[TMP48]], align 4
// CHECK-USE-I386-NEXT: [[TMP49:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK-USE-I386-NEXT: [[TMP49:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6
// CHECK-USE-I386-NEXT: store i64 4, i64* [[TMP49]], align 4 // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP49]], align 4
// CHECK-USE-I386-NEXT: [[TMP50:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5 // CHECK-USE-I386-NEXT: [[TMP50:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP50]], align 4 // CHECK-USE-I386-NEXT: [[TMP51:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP51:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK-USE-I386-NEXT: [[TMP52:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l606.region_id, i32 7, i8** [[TMP50]], i8** [[TMP51]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null)
// CHECK-USE-I386-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to %struct.ST** // CHECK-USE-I386-NEXT: [[TMP53:%.*]] = icmp ne i32 [[TMP52]], 0
// CHECK-USE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP52]], align 4 // CHECK-USE-I386-NEXT: br i1 [[TMP53]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-USE-I386-NEXT: [[TMP53:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK-USE-I386-NEXT: [[TMP54:%.*]] = bitcast i8** [[TMP53]] to i32**
// CHECK-USE-I386-NEXT: store i32* [[J2]], i32** [[TMP54]], align 4
// CHECK-USE-I386-NEXT: [[TMP55:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6
// CHECK-USE-I386-NEXT: store i64 4, i64* [[TMP55]], align 4
// CHECK-USE-I386-NEXT: [[TMP56:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP56]], align 4
// CHECK-USE-I386-NEXT: [[TMP57:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP58:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP59:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP60:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l654.region_id, i32 7, i8** [[TMP57]], i8** [[TMP58]], i64* [[TMP59]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null)
// CHECK-USE-I386-NEXT: [[TMP61:%.*]] = icmp ne i32 [[TMP60]], 0
// CHECK-USE-I386-NEXT: br i1 [[TMP61]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-USE-I386: omp_offload.failed: // CHECK-USE-I386: omp_offload.failed:
// CHECK-USE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l654(%struct.ST* [[ST1]], i32* [[A]], %struct.ST* [[ST2]]) #[[ATTR2:[0-9]+]] // CHECK-USE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l606(%struct.ST* [[ST1]], i32* [[A]], %struct.ST* [[ST2]]) #[[ATTR2:[0-9]+]]
// CHECK-USE-I386-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-USE-I386-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK-USE-I386: omp_offload.cont: // CHECK-USE-I386: omp_offload.cont:
// CHECK-USE-I386-NEXT: [[TMP62:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[TMP54:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32** // CHECK-USE-I386-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to i32**
// CHECK-USE-I386-NEXT: store i32* [[A]], i32** [[TMP63]], align 4 // CHECK-USE-I386-NEXT: store i32* [[A]], i32** [[TMP55]], align 4
// CHECK-USE-I386-NEXT: [[TMP64:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[TMP56:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32** // CHECK-USE-I386-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to i32**
// CHECK-USE-I386-NEXT: store i32* [[A]], i32** [[TMP65]], align 4 // CHECK-USE-I386-NEXT: store i32* [[A]], i32** [[TMP57]], align 4
// CHECK-USE-I386-NEXT: [[TMP66:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[TMP58:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i32 0, i32 0
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP66]], align 4 // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP58]], align 4
// CHECK-USE-I386-NEXT: [[TMP67:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[TMP59:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP68:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[TMP60:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP69:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l668.region_id, i32 1, i8** [[TMP67]], i8** [[TMP68]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.1, i32 0, i32 0), i8** null, i8** null) // CHECK-USE-I386-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l620.region_id, i32 1, i8** [[TMP59]], i8** [[TMP60]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.1, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null)
// CHECK-USE-I386-NEXT: [[TMP70:%.*]] = icmp ne i32 [[TMP69]], 0 // CHECK-USE-I386-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0
// CHECK-USE-I386-NEXT: br i1 [[TMP70]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]] // CHECK-USE-I386-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
// CHECK-USE-I386: omp_offload.failed6: // CHECK-USE-I386: omp_offload.failed6:
// CHECK-USE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l668(i32* [[A]]) #[[ATTR2]] // CHECK-USE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l620(i32* [[A]]) #[[ATTR2]]
// CHECK-USE-I386-NEXT: br label [[OMP_OFFLOAD_CONT7]] // CHECK-USE-I386-NEXT: br label [[OMP_OFFLOAD_CONT7]]
// CHECK-USE-I386: omp_offload.cont7: // CHECK-USE-I386: omp_offload.cont7:
// CHECK-USE-I386-NEXT: ret void // CHECK-USE-I386-NEXT: ret void
// //
// CHECK-NOUSE-PPC64LE-LABEL: @_Z20explicit_maps_singlei( // CHECK-NOUSE-PPC64LE-LABEL: @_Z20explicit_maps_singlei(
// CHECK-NOUSE-PPC64LE-NEXT: entry: // CHECK-NOUSE-PPC64LE-NEXT: entry:
// CHECK-NOUSE-PPC64LE-NEXT: [[II_ADDR:%.*]] = alloca i32, align 4 // CHECK-NOUSE-PPC64LE-NEXT: [[II_ADDR:%.*]] = alloca i32, align 4
// CHECK-NOUSE-PPC64LE-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK-NOUSE-PPC64LE-NEXT: [[A:%.*]] = alloca i32, align 4
// CHECK-NOUSE-PPC64LE-NEXT: [[ST1:%.*]] = alloca [[STRUCT_ST:%.*]], align 4 // CHECK-NOUSE-PPC64LE-NEXT: [[ST1:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK-NOUSE-PPC64LE-NEXT: [[ST2:%.*]] = alloca [[STRUCT_ST]], align 4 // CHECK-NOUSE-PPC64LE-NEXT: [[ST2:%.*]] = alloca [[STRUCT_ST]], align 4
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [7 x i8*], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [7 x i8*], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [7 x i8*], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [7 x i8*], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [7 x i8*], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [7 x i8*], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [7 x i64], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x i8*], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x i8*], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x i8*], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x i8*], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x i8*], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x i8*], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i32 [[II:%.*]], i32* [[II_ADDR]], align 4 // CHECK-NOUSE-PPC64LE-NEXT: store i32 [[II:%.*]], i32* [[II_ADDR]], align 4
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP0:%.*]] = load i32, i32* [[II_ADDR]], align 4 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP0:%.*]] = load i32, i32* [[II_ADDR]], align 4
// CHECK-NOUSE-PPC64LE-NEXT: store i32 [[TMP0]], i32* [[A]], align 4 // CHECK-NOUSE-PPC64LE-NEXT: store i32 [[TMP0]], i32* [[A]], align 4
// CHECK-NOUSE-PPC64LE-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 1 // CHECK-NOUSE-PPC64LE-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 1
Show All 14 Lines
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]] // CHECK-NOUSE-PPC64LE-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]]
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK-NOUSE-PPC64LE-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to %struct.ST** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to %struct.ST**
// CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP16]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP16]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP18]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP18]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 0), align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i64 [[TMP7]], i64* [[TMP19]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP20:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP20]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP20:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.ST**
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.ST** // CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP21]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP22]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP22:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP23]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP24]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP24:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i64 4, i64* [[TMP25]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to %struct.ST**
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP26]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to %struct.ST** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP28]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[J]], i32** [[TMP28]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[J]], i32** [[TMP30]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP30:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP31:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: store i64 4, i64* [[TMP31]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP31]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP32]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP33:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP33]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP34:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP34]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to %struct.ST**
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP36]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP36]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP37:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP37:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK-NOUSE-PPC64LE-NEXT: store i64 4, i64* [[TMP37]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP38:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I1]], i32** [[TMP38]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP38]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i64 [[TMP14]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP39:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP39:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to %struct.ST** // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP40]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP40:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP41:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to %struct.ST**
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP42:%.*]] = bitcast i8** [[TMP41]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP41]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I1]], i32** [[TMP42]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP42:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP43:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: store i64 [[TMP14]], i64* [[TMP43]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I1]], i32** [[TMP43]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP44:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP44:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP44]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP44]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP45:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP45:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to %struct.ST** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to %struct.ST**
// CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP46]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP46]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP47:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP47:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I1]], i32** [[TMP48]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[J2]], i32** [[TMP48]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP49:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP49:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6
// CHECK-NOUSE-PPC64LE-NEXT: store i64 4, i64* [[TMP49]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP49]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP50:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP50:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP50]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP51:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP51:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP52:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l606.region_id, i32 7, i8** [[TMP50]], i8** [[TMP51]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null)
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to %struct.ST** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP53:%.*]] = icmp ne i32 [[TMP52]], 0
// CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP52]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: br i1 [[TMP53]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP53:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP54:%.*]] = bitcast i8** [[TMP53]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[J2]], i32** [[TMP54]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP55:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6
// CHECK-NOUSE-PPC64LE-NEXT: store i64 4, i64* [[TMP55]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP56:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP56]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP57:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP58:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP59:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP60:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l654.region_id, i32 7, i8** [[TMP57]], i8** [[TMP58]], i64* [[TMP59]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null)
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP61:%.*]] = icmp ne i32 [[TMP60]], 0
// CHECK-NOUSE-PPC64LE-NEXT: br i1 [[TMP61]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-NOUSE-PPC64LE: omp_offload.failed: // CHECK-NOUSE-PPC64LE: omp_offload.failed:
// CHECK-NOUSE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l654() #[[ATTR2:[0-9]+]] // CHECK-NOUSE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l606() #[[ATTR2:[0-9]+]]
// CHECK-NOUSE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-NOUSE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK-NOUSE-PPC64LE: omp_offload.cont: // CHECK-NOUSE-PPC64LE: omp_offload.cont:
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP62:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP54:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP63]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP55]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP64:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP56:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP65]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[A]], i32** [[TMP57]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP66:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP58:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP66]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP58]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP67:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP59:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP68:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP60:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP69:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l668.region_id, i32 1, i8** [[TMP67]], i8** [[TMP68]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.1, i32 0, i32 0), i8** null, i8** null) // CHECK-NOUSE-PPC64LE-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l620.region_id, i32 1, i8** [[TMP59]], i8** [[TMP60]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.1, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null)
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP70:%.*]] = icmp ne i32 [[TMP69]], 0 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0
// CHECK-NOUSE-PPC64LE-NEXT: br i1 [[TMP70]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]] // CHECK-NOUSE-PPC64LE-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
// CHECK-NOUSE-PPC64LE: omp_offload.failed6: // CHECK-NOUSE-PPC64LE: omp_offload.failed6:
// CHECK-NOUSE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l668() #[[ATTR2]] // CHECK-NOUSE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l620() #[[ATTR2]]
// CHECK-NOUSE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT7]] // CHECK-NOUSE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT7]]
// CHECK-NOUSE-PPC64LE: omp_offload.cont7: // CHECK-NOUSE-PPC64LE: omp_offload.cont7:
// CHECK-NOUSE-PPC64LE-NEXT: ret void // CHECK-NOUSE-PPC64LE-NEXT: ret void
// //
// CHECK-NOUSE-I386-LABEL: @_Z20explicit_maps_singlei( // CHECK-NOUSE-I386-LABEL: @_Z20explicit_maps_singlei(
// CHECK-NOUSE-I386-NEXT: entry: // CHECK-NOUSE-I386-NEXT: entry:
// CHECK-NOUSE-I386-NEXT: [[II_ADDR:%.*]] = alloca i32, align 4 // CHECK-NOUSE-I386-NEXT: [[II_ADDR:%.*]] = alloca i32, align 4
// CHECK-NOUSE-I386-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK-NOUSE-I386-NEXT: [[A:%.*]] = alloca i32, align 4
// CHECK-NOUSE-I386-NEXT: [[ST1:%.*]] = alloca [[STRUCT_ST:%.*]], align 4 // CHECK-NOUSE-I386-NEXT: [[ST1:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK-NOUSE-I386-NEXT: [[ST2:%.*]] = alloca [[STRUCT_ST]], align 4 // CHECK-NOUSE-I386-NEXT: [[ST2:%.*]] = alloca [[STRUCT_ST]], align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [7 x i8*], align 4 // CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [7 x i8*], align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [7 x i8*], align 4 // CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [7 x i8*], align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [7 x i8*], align 4 // CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [7 x i8*], align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [7 x i64], align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x i8*], align 4 // CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x i8*], align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x i8*], align 4 // CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x i8*], align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x i8*], align 4 // CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x i8*], align 4
// CHECK-NOUSE-I386-NEXT: store i32 [[II:%.*]], i32* [[II_ADDR]], align 4 // CHECK-NOUSE-I386-NEXT: store i32 [[II:%.*]], i32* [[II_ADDR]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP0:%.*]] = load i32, i32* [[II_ADDR]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP0:%.*]] = load i32, i32* [[II_ADDR]], align 4
// CHECK-NOUSE-I386-NEXT: store i32 [[TMP0]], i32* [[A]], align 4 // CHECK-NOUSE-I386-NEXT: store i32 [[TMP0]], i32* [[A]], align 4
// CHECK-NOUSE-I386-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 1 // CHECK-NOUSE-I386-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[ST1]], i32 0, i32 1
Show All 14 Lines
// CHECK-NOUSE-I386-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]] // CHECK-NOUSE-I386-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]]
// CHECK-NOUSE-I386-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK-NOUSE-I386-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK-NOUSE-I386-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[TMP15:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to %struct.ST** // CHECK-NOUSE-I386-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to %struct.ST**
// CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP16]], align 4 // CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP16]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[TMP17:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** // CHECK-NOUSE-I386-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32**
// CHECK-NOUSE-I386-NEXT: store i32* [[I]], i32** [[TMP18]], align 4 // CHECK-NOUSE-I386-NEXT: store i32* [[I]], i32** [[TMP18]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 0), align 4
// CHECK-NOUSE-I386-NEXT: store i64 [[TMP7]], i64* [[TMP19]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP20:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP20]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP20:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK-NOUSE-I386-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK-NOUSE-I386-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.ST**
// CHECK-NOUSE-I386-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.ST** // CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP21]], align 4
// CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP22]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP22:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK-NOUSE-I386-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK-NOUSE-I386-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32**
// CHECK-NOUSE-I386-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32** // CHECK-NOUSE-I386-NEXT: store i32* [[I]], i32** [[TMP23]], align 4
// CHECK-NOUSE-I386-NEXT: store i32* [[I]], i32** [[TMP24]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP24:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK-NOUSE-I386-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK-NOUSE-I386-NEXT: store i64 4, i64* [[TMP25]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK-NOUSE-I386-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK-NOUSE-I386-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to %struct.ST**
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP26]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK-NOUSE-I386-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK-NOUSE-I386-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to %struct.ST** // CHECK-NOUSE-I386-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32**
// CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST1]], %struct.ST** [[TMP28]], align 4 // CHECK-NOUSE-I386-NEXT: store i32* [[J]], i32** [[TMP28]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK-NOUSE-I386-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK-NOUSE-I386-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32** // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK-NOUSE-I386-NEXT: store i32* [[J]], i32** [[TMP30]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP30:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK-NOUSE-I386-NEXT: [[TMP31:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK-NOUSE-I386-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32**
// CHECK-NOUSE-I386-NEXT: store i64 4, i64* [[TMP31]], align 4 // CHECK-NOUSE-I386-NEXT: store i32* [[A]], i32** [[TMP31]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK-NOUSE-I386-NEXT: [[TMP32:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP32]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK-NOUSE-I386-NEXT: [[TMP33:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK-NOUSE-I386-NEXT: store i32* [[A]], i32** [[TMP33]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32** // CHECK-NOUSE-I386-NEXT: [[TMP34:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK-NOUSE-I386-NEXT: store i32* [[A]], i32** [[TMP34]], align 4 // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK-NOUSE-I386-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK-NOUSE-I386-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK-NOUSE-I386-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to %struct.ST**
// CHECK-NOUSE-I386-NEXT: store i32* [[A]], i32** [[TMP36]], align 4 // CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP36]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP37:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK-NOUSE-I386-NEXT: [[TMP37:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK-NOUSE-I386-NEXT: store i64 4, i64* [[TMP37]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32**
// CHECK-NOUSE-I386-NEXT: [[TMP38:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK-NOUSE-I386-NEXT: store i32* [[I1]], i32** [[TMP38]], align 4
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP38]], align 4 // CHECK-NOUSE-I386-NEXT: store i64 [[TMP14]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK-NOUSE-I386-NEXT: [[TMP39:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK-NOUSE-I386-NEXT: [[TMP39:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK-NOUSE-I386-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to %struct.ST** // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP40]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP40:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK-NOUSE-I386-NEXT: [[TMP41:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK-NOUSE-I386-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to %struct.ST**
// CHECK-NOUSE-I386-NEXT: [[TMP42:%.*]] = bitcast i8** [[TMP41]] to i32** // CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP41]], align 4
// CHECK-NOUSE-I386-NEXT: store i32* [[I1]], i32** [[TMP42]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP42:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK-NOUSE-I386-NEXT: [[TMP43:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK-NOUSE-I386-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK-NOUSE-I386-NEXT: store i64 [[TMP14]], i64* [[TMP43]], align 4 // CHECK-NOUSE-I386-NEXT: store i32* [[I1]], i32** [[TMP43]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP44:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK-NOUSE-I386-NEXT: [[TMP44:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP44]], align 4 // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP44]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP45:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK-NOUSE-I386-NEXT: [[TMP45:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK-NOUSE-I386-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to %struct.ST** // CHECK-NOUSE-I386-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to %struct.ST**
// CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP46]], align 4 // CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP46]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP47:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK-NOUSE-I386-NEXT: [[TMP47:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK-NOUSE-I386-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32** // CHECK-NOUSE-I386-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i32**
// CHECK-NOUSE-I386-NEXT: store i32* [[I1]], i32** [[TMP48]], align 4 // CHECK-NOUSE-I386-NEXT: store i32* [[J2]], i32** [[TMP48]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP49:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK-NOUSE-I386-NEXT: [[TMP49:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6
// CHECK-NOUSE-I386-NEXT: store i64 4, i64* [[TMP49]], align 4 // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP49]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP50:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5 // CHECK-NOUSE-I386-NEXT: [[TMP50:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP50]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP51:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP51:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK-NOUSE-I386-NEXT: [[TMP52:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l606.region_id, i32 7, i8** [[TMP50]], i8** [[TMP51]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null)
// CHECK-NOUSE-I386-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to %struct.ST** // CHECK-NOUSE-I386-NEXT: [[TMP53:%.*]] = icmp ne i32 [[TMP52]], 0
// CHECK-NOUSE-I386-NEXT: store %struct.ST* [[ST2]], %struct.ST** [[TMP52]], align 4 // CHECK-NOUSE-I386-NEXT: br i1 [[TMP53]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-NOUSE-I386-NEXT: [[TMP53:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK-NOUSE-I386-NEXT: [[TMP54:%.*]] = bitcast i8** [[TMP53]] to i32**
// CHECK-NOUSE-I386-NEXT: store i32* [[J2]], i32** [[TMP54]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP55:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6
// CHECK-NOUSE-I386-NEXT: store i64 4, i64* [[TMP55]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP56:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP56]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP57:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP58:%.*]] = getelementptr inbounds [7 x i8*], [7 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP59:%.*]] = getelementptr inbounds [7 x i64], [7 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP60:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l654.region_id, i32 7, i8** [[TMP57]], i8** [[TMP58]], i64* [[TMP59]], i64* getelementptr inbounds ([7 x i64], [7 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null)
// CHECK-NOUSE-I386-NEXT: [[TMP61:%.*]] = icmp ne i32 [[TMP60]], 0
// CHECK-NOUSE-I386-NEXT: br i1 [[TMP61]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-NOUSE-I386: omp_offload.failed: // CHECK-NOUSE-I386: omp_offload.failed:
// CHECK-NOUSE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l654() #[[ATTR2:[0-9]+]] // CHECK-NOUSE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l606() #[[ATTR2:[0-9]+]]
// CHECK-NOUSE-I386-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-NOUSE-I386-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK-NOUSE-I386: omp_offload.cont: // CHECK-NOUSE-I386: omp_offload.cont:
// CHECK-NOUSE-I386-NEXT: [[TMP62:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[TMP54:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32** // CHECK-NOUSE-I386-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to i32**
// CHECK-NOUSE-I386-NEXT: store i32* [[A]], i32** [[TMP63]], align 4 // CHECK-NOUSE-I386-NEXT: store i32* [[A]], i32** [[TMP55]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP64:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[TMP56:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32** // CHECK-NOUSE-I386-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to i32**
// CHECK-NOUSE-I386-NEXT: store i32* [[A]], i32** [[TMP65]], align 4 // CHECK-NOUSE-I386-NEXT: store i32* [[A]], i32** [[TMP57]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP66:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[TMP58:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP66]], align 4 // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP58]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP67:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[TMP59:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP68:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[TMP60:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP69:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l668.region_id, i32 1, i8** [[TMP67]], i8** [[TMP68]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.1, i32 0, i32 0), i8** null, i8** null) // CHECK-NOUSE-I386-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l620.region_id, i32 1, i8** [[TMP59]], i8** [[TMP60]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.1, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null)
// CHECK-NOUSE-I386-NEXT: [[TMP70:%.*]] = icmp ne i32 [[TMP69]], 0 // CHECK-NOUSE-I386-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0
// CHECK-NOUSE-I386-NEXT: br i1 [[TMP70]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]] // CHECK-NOUSE-I386-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
// CHECK-NOUSE-I386: omp_offload.failed6: // CHECK-NOUSE-I386: omp_offload.failed6:
// CHECK-NOUSE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l668() #[[ATTR2]] // CHECK-NOUSE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20explicit_maps_singlei_l620() #[[ATTR2]]
// CHECK-NOUSE-I386-NEXT: br label [[OMP_OFFLOAD_CONT7]] // CHECK-NOUSE-I386-NEXT: br label [[OMP_OFFLOAD_CONT7]]
// CHECK-NOUSE-I386: omp_offload.cont7: // CHECK-NOUSE-I386: omp_offload.cont7:
// CHECK-NOUSE-I386-NEXT: ret void // CHECK-NOUSE-I386-NEXT: ret void
// //
void explicit_maps_single(int ii) { void explicit_maps_single(int ii) {
// Map of a scalar. // Map of a scalar.
int a = ii; int a = ii;
Show All 26 Lines
} }
// CHECK-USE-PPC64LE-LABEL: @_ZN2ST20test_present_membersEv( // CHECK-USE-PPC64LE-LABEL: @_ZN2ST20test_present_membersEv(
// CHECK-USE-PPC64LE-NEXT: entry: // CHECK-USE-PPC64LE-NEXT: entry:
// CHECK-USE-PPC64LE-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.ST*, align 8 // CHECK-USE-PPC64LE-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.ST*, align 8
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK-USE-PPC64LE-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[THIS:%.*]], %struct.ST** [[THIS_ADDR]], align 8 // CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[THIS:%.*]], %struct.ST** [[THIS_ADDR]], align 8
// CHECK-USE-PPC64LE-NEXT: [[THIS1:%.*]] = load %struct.ST*, %struct.ST** [[THIS_ADDR]], align 8 // CHECK-USE-PPC64LE-NEXT: [[THIS1:%.*]] = load %struct.ST*, %struct.ST** [[THIS_ADDR]], align 8
// CHECK-USE-PPC64LE-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.ST* [[THIS1]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.ST* [[THIS1]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[THIS1]], i32 0, i32 1 // CHECK-USE-PPC64LE-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[THIS1]], i32 0, i32 1
// CHECK-USE-PPC64LE-NEXT: [[TMP0:%.*]] = getelementptr i32, i32* [[J]], i32 1 // CHECK-USE-PPC64LE-NEXT: [[TMP0:%.*]] = getelementptr i32, i32* [[J]], i32 1
// CHECK-USE-PPC64LE-NEXT: [[TMP1:%.*]] = bitcast i32* [[I]] to i8* // CHECK-USE-PPC64LE-NEXT: [[TMP1:%.*]] = bitcast i32* [[I]] to i8*
// CHECK-USE-PPC64LE-NEXT: [[TMP2:%.*]] = bitcast i32* [[TMP0]] to i8* // CHECK-USE-PPC64LE-NEXT: [[TMP2:%.*]] = bitcast i32* [[TMP0]] to i8*
// CHECK-USE-PPC64LE-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK-USE-PPC64LE-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK-USE-PPC64LE-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK-USE-PPC64LE-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK-USE-PPC64LE-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK-USE-PPC64LE-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK-USE-PPC64LE-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK-USE-PPC64LE-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK-USE-PPC64LE-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.ST** // CHECK-USE-PPC64LE-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.ST**
// CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP8]], align 8 // CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP8]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32** // CHECK-USE-PPC64LE-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32**
// CHECK-USE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP10]], align 8 // CHECK-USE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP10]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-USE-PPC64LE-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.3, i32 0, i32 0), align 8
// CHECK-USE-PPC64LE-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK-USE-PPC64LE-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP12]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK-USE-PPC64LE-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK-USE-PPC64LE-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.ST**
// CHECK-USE-PPC64LE-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.ST** // CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP13]], align 8
// CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP14]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK-USE-PPC64LE-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK-USE-PPC64LE-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32**
// CHECK-USE-PPC64LE-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32** // CHECK-USE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP15]], align 8
// CHECK-USE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP16]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK-USE-PPC64LE-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK-USE-PPC64LE-NEXT: store i64 4, i64* [[TMP17]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK-USE-PPC64LE-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK-USE-PPC64LE-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.ST**
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP18]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK-USE-PPC64LE-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK-USE-PPC64LE-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.ST** // CHECK-USE-PPC64LE-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32**
// CHECK-USE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP20]], align 8 // CHECK-USE-PPC64LE-NEXT: store i32* [[J]], i32** [[TMP20]], align 8
// CHECK-USE-PPC64LE-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK-USE-PPC64LE-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK-USE-PPC64LE-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32** // CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK-USE-PPC64LE-NEXT: store i32* [[J]], i32** [[TMP22]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK-USE-PPC64LE-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: store i64 4, i64* [[TMP23]], align 8 // CHECK-USE-PPC64LE-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l843.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.3, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null)
// CHECK-USE-PPC64LE-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK-USE-PPC64LE-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK-USE-PPC64LE-NEXT: store i8* null, i8** [[TMP24]], align 8 // CHECK-USE-PPC64LE-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-USE-PPC64LE-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK-USE-PPC64LE-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l919.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null)
// CHECK-USE-PPC64LE-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
// CHECK-USE-PPC64LE-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-USE-PPC64LE: omp_offload.failed: // CHECK-USE-PPC64LE: omp_offload.failed:
// CHECK-USE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l919(%struct.ST* [[THIS1]]) #[[ATTR2]] // CHECK-USE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l843(%struct.ST* [[THIS1]]) #[[ATTR2]]
// CHECK-USE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-USE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK-USE-PPC64LE: omp_offload.cont: // CHECK-USE-PPC64LE: omp_offload.cont:
// CHECK-USE-PPC64LE-NEXT: ret void // CHECK-USE-PPC64LE-NEXT: ret void
// //
// CHECK-USE-I386-LABEL: @_ZN2ST20test_present_membersEv( // CHECK-USE-I386-LABEL: @_ZN2ST20test_present_membersEv(
// CHECK-USE-I386-NEXT: entry: // CHECK-USE-I386-NEXT: entry:
// CHECK-USE-I386-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.ST*, align 4 // CHECK-USE-I386-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.ST*, align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK-USE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK-USE-I386-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK-USE-I386-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK-USE-I386-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK-USE-I386-NEXT: store %struct.ST* [[THIS:%.*]], %struct.ST** [[THIS_ADDR]], align 4 // CHECK-USE-I386-NEXT: store %struct.ST* [[THIS:%.*]], %struct.ST** [[THIS_ADDR]], align 4
// CHECK-USE-I386-NEXT: [[THIS1:%.*]] = load %struct.ST*, %struct.ST** [[THIS_ADDR]], align 4 // CHECK-USE-I386-NEXT: [[THIS1:%.*]] = load %struct.ST*, %struct.ST** [[THIS_ADDR]], align 4
// CHECK-USE-I386-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.ST* [[THIS1]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.ST* [[THIS1]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[THIS1]], i32 0, i32 1 // CHECK-USE-I386-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[THIS1]], i32 0, i32 1
// CHECK-USE-I386-NEXT: [[TMP0:%.*]] = getelementptr i32, i32* [[J]], i32 1 // CHECK-USE-I386-NEXT: [[TMP0:%.*]] = getelementptr i32, i32* [[J]], i32 1
// CHECK-USE-I386-NEXT: [[TMP1:%.*]] = bitcast i32* [[I]] to i8* // CHECK-USE-I386-NEXT: [[TMP1:%.*]] = bitcast i32* [[I]] to i8*
// CHECK-USE-I386-NEXT: [[TMP2:%.*]] = bitcast i32* [[TMP0]] to i8* // CHECK-USE-I386-NEXT: [[TMP2:%.*]] = bitcast i32* [[TMP0]] to i8*
// CHECK-USE-I386-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK-USE-I386-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK-USE-I386-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK-USE-I386-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK-USE-I386-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK-USE-I386-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK-USE-I386-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK-USE-I386-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK-USE-I386-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.ST** // CHECK-USE-I386-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.ST**
// CHECK-USE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP8]], align 4 // CHECK-USE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP8]], align 4
// CHECK-USE-I386-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK-USE-I386-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32** // CHECK-USE-I386-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32**
// CHECK-USE-I386-NEXT: store i32* [[I]], i32** [[TMP10]], align 4 // CHECK-USE-I386-NEXT: store i32* [[I]], i32** [[TMP10]], align 4
// CHECK-USE-I386-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-USE-I386-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.3, i32 0, i32 0), align 4
// CHECK-USE-I386-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 4 // CHECK-USE-I386-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP11]], align 4
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP12]], align 4 // CHECK-USE-I386-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK-USE-I386-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK-USE-I386-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.ST**
// CHECK-USE-I386-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.ST** // CHECK-USE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP13]], align 4
// CHECK-USE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP14]], align 4 // CHECK-USE-I386-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK-USE-I386-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK-USE-I386-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32**
// CHECK-USE-I386-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32** // CHECK-USE-I386-NEXT: store i32* [[I]], i32** [[TMP15]], align 4
// CHECK-USE-I386-NEXT: store i32* [[I]], i32** [[TMP16]], align 4 // CHECK-USE-I386-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK-USE-I386-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP16]], align 4
// CHECK-USE-I386-NEXT: store i64 4, i64* [[TMP17]], align 4 // CHECK-USE-I386-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK-USE-I386-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK-USE-I386-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.ST**
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK-USE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP18]], align 4
// CHECK-USE-I386-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK-USE-I386-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK-USE-I386-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.ST** // CHECK-USE-I386-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32**
// CHECK-USE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP20]], align 4 // CHECK-USE-I386-NEXT: store i32* [[J]], i32** [[TMP20]], align 4
// CHECK-USE-I386-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK-USE-I386-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK-USE-I386-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32** // CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP21]], align 4
// CHECK-USE-I386-NEXT: store i32* [[J]], i32** [[TMP22]], align 4 // CHECK-USE-I386-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK-USE-I386-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: store i64 4, i64* [[TMP23]], align 4 // CHECK-USE-I386-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l843.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.3, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null)
// CHECK-USE-I386-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK-USE-I386-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK-USE-I386-NEXT: store i8* null, i8** [[TMP24]], align 4 // CHECK-USE-I386-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-USE-I386-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK-USE-I386-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l919.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null)
// CHECK-USE-I386-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
// CHECK-USE-I386-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-USE-I386: omp_offload.failed: // CHECK-USE-I386: omp_offload.failed:
// CHECK-USE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l919(%struct.ST* [[THIS1]]) #[[ATTR2]] // CHECK-USE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l843(%struct.ST* [[THIS1]]) #[[ATTR2]]
// CHECK-USE-I386-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-USE-I386-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK-USE-I386: omp_offload.cont: // CHECK-USE-I386: omp_offload.cont:
// CHECK-USE-I386-NEXT: ret void // CHECK-USE-I386-NEXT: ret void
// //
// CHECK-NOUSE-PPC64LE-LABEL: @_ZN2ST20test_present_membersEv( // CHECK-NOUSE-PPC64LE-LABEL: @_ZN2ST20test_present_membersEv(
// CHECK-NOUSE-PPC64LE-NEXT: entry: // CHECK-NOUSE-PPC64LE-NEXT: entry:
// CHECK-NOUSE-PPC64LE-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.ST*, align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.ST*, align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[THIS:%.*]], %struct.ST** [[THIS_ADDR]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[THIS:%.*]], %struct.ST** [[THIS_ADDR]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[THIS1:%.*]] = load %struct.ST*, %struct.ST** [[THIS_ADDR]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[THIS1:%.*]] = load %struct.ST*, %struct.ST** [[THIS_ADDR]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.ST* [[THIS1]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.ST* [[THIS1]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[THIS1]], i32 0, i32 1 // CHECK-NOUSE-PPC64LE-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[THIS1]], i32 0, i32 1
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP0:%.*]] = getelementptr i32, i32* [[J]], i32 1 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP0:%.*]] = getelementptr i32, i32* [[J]], i32 1
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP1:%.*]] = bitcast i32* [[I]] to i8* // CHECK-NOUSE-PPC64LE-NEXT: [[TMP1:%.*]] = bitcast i32* [[I]] to i8*
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP2:%.*]] = bitcast i32* [[TMP0]] to i8* // CHECK-NOUSE-PPC64LE-NEXT: [[TMP2:%.*]] = bitcast i32* [[TMP0]] to i8*
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK-NOUSE-PPC64LE-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK-NOUSE-PPC64LE-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.ST** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.ST**
// CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP8]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP8]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP10]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP10]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-NOUSE-PPC64LE-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.3, i32 0, i32 0), align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP12]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.ST**
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.ST** // CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP13]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP14]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP15]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[I]], i32** [[TMP16]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i64 4, i64* [[TMP17]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.ST**
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP18]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.ST** // CHECK-NOUSE-PPC64LE-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32**
// CHECK-NOUSE-PPC64LE-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP20]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: store i32* [[J]], i32** [[TMP20]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32** // CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK-NOUSE-PPC64LE-NEXT: store i32* [[J]], i32** [[TMP22]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: store i64 4, i64* [[TMP23]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l843.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.3, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null)
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK-NOUSE-PPC64LE-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK-NOUSE-PPC64LE-NEXT: store i8* null, i8** [[TMP24]], align 8 // CHECK-NOUSE-PPC64LE-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l919.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null)
// CHECK-NOUSE-PPC64LE-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
// CHECK-NOUSE-PPC64LE-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-NOUSE-PPC64LE: omp_offload.failed: // CHECK-NOUSE-PPC64LE: omp_offload.failed:
// CHECK-NOUSE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l919() #[[ATTR2]] // CHECK-NOUSE-PPC64LE-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l843() #[[ATTR2]]
// CHECK-NOUSE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-NOUSE-PPC64LE-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK-NOUSE-PPC64LE: omp_offload.cont: // CHECK-NOUSE-PPC64LE: omp_offload.cont:
// CHECK-NOUSE-PPC64LE-NEXT: ret void // CHECK-NOUSE-PPC64LE-NEXT: ret void
// //
// CHECK-NOUSE-I386-LABEL: @_ZN2ST20test_present_membersEv( // CHECK-NOUSE-I386-LABEL: @_ZN2ST20test_present_membersEv(
// CHECK-NOUSE-I386-NEXT: entry: // CHECK-NOUSE-I386-NEXT: entry:
// CHECK-NOUSE-I386-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.ST*, align 4 // CHECK-NOUSE-I386-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.ST*, align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK-NOUSE-I386-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK-NOUSE-I386-NEXT: store %struct.ST* [[THIS:%.*]], %struct.ST** [[THIS_ADDR]], align 4 // CHECK-NOUSE-I386-NEXT: store %struct.ST* [[THIS:%.*]], %struct.ST** [[THIS_ADDR]], align 4
// CHECK-NOUSE-I386-NEXT: [[THIS1:%.*]] = load %struct.ST*, %struct.ST** [[THIS_ADDR]], align 4 // CHECK-NOUSE-I386-NEXT: [[THIS1:%.*]] = load %struct.ST*, %struct.ST** [[THIS_ADDR]], align 4
// CHECK-NOUSE-I386-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.ST* [[THIS1]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[I:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.ST* [[THIS1]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[THIS1]], i32 0, i32 1 // CHECK-NOUSE-I386-NEXT: [[J:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.ST* [[THIS1]], i32 0, i32 1
// CHECK-NOUSE-I386-NEXT: [[TMP0:%.*]] = getelementptr i32, i32* [[J]], i32 1 // CHECK-NOUSE-I386-NEXT: [[TMP0:%.*]] = getelementptr i32, i32* [[J]], i32 1
// CHECK-NOUSE-I386-NEXT: [[TMP1:%.*]] = bitcast i32* [[I]] to i8* // CHECK-NOUSE-I386-NEXT: [[TMP1:%.*]] = bitcast i32* [[I]] to i8*
// CHECK-NOUSE-I386-NEXT: [[TMP2:%.*]] = bitcast i32* [[TMP0]] to i8* // CHECK-NOUSE-I386-NEXT: [[TMP2:%.*]] = bitcast i32* [[TMP0]] to i8*
// CHECK-NOUSE-I386-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK-NOUSE-I386-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK-NOUSE-I386-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK-NOUSE-I386-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK-NOUSE-I386-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK-NOUSE-I386-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK-NOUSE-I386-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK-NOUSE-I386-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK-NOUSE-I386-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.ST** // CHECK-NOUSE-I386-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.ST**
// CHECK-NOUSE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP8]], align 4 // CHECK-NOUSE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP8]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32** // CHECK-NOUSE-I386-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32**
// CHECK-NOUSE-I386-NEXT: store i32* [[I]], i32** [[TMP10]], align 4 // CHECK-NOUSE-I386-NEXT: store i32* [[I]], i32** [[TMP10]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.3, i32 0, i32 0), align 4
// CHECK-NOUSE-I386-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP11]], align 4
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP12]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK-NOUSE-I386-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK-NOUSE-I386-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.ST**
// CHECK-NOUSE-I386-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.ST** // CHECK-NOUSE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP13]], align 4
// CHECK-NOUSE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP14]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK-NOUSE-I386-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK-NOUSE-I386-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32**
// CHECK-NOUSE-I386-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32** // CHECK-NOUSE-I386-NEXT: store i32* [[I]], i32** [[TMP15]], align 4
// CHECK-NOUSE-I386-NEXT: store i32* [[I]], i32** [[TMP16]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK-NOUSE-I386-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP16]], align 4
// CHECK-NOUSE-I386-NEXT: store i64 4, i64* [[TMP17]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK-NOUSE-I386-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK-NOUSE-I386-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.ST**
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK-NOUSE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP18]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK-NOUSE-I386-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK-NOUSE-I386-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.ST** // CHECK-NOUSE-I386-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32**
// CHECK-NOUSE-I386-NEXT: store %struct.ST* [[THIS1]], %struct.ST** [[TMP20]], align 4 // CHECK-NOUSE-I386-NEXT: store i32* [[J]], i32** [[TMP20]], align 4
// CHECK-NOUSE-I386-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK-NOUSE-I386-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK-NOUSE-I386-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32** // CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP21]], align 4
// CHECK-NOUSE-I386-NEXT: store i32* [[J]], i32** [[TMP22]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK-NOUSE-I386-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: store i64 4, i64* [[TMP23]], align 4 // CHECK-NOUSE-I386-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l843.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.3, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null)
// CHECK-NOUSE-I386-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK-NOUSE-I386-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK-NOUSE-I386-NEXT: store i8* null, i8** [[TMP24]], align 4 // CHECK-NOUSE-I386-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-NOUSE-I386-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK-NOUSE-I386-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l919.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null)
// CHECK-NOUSE-I386-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
// CHECK-NOUSE-I386-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK-NOUSE-I386: omp_offload.failed: // CHECK-NOUSE-I386: omp_offload.failed:
// CHECK-NOUSE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l919() #[[ATTR2]] // CHECK-NOUSE-I386-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2ST20test_present_membersEv_l843() #[[ATTR2]]
// CHECK-NOUSE-I386-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-NOUSE-I386-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK-NOUSE-I386: omp_offload.cont: // CHECK-NOUSE-I386: omp_offload.cont:
// CHECK-NOUSE-I386-NEXT: ret void // CHECK-NOUSE-I386-NEXT: ret void
// //
void ST::test_present_members() { void ST::test_present_members() {
// Make sure the struct picks up ompx_hold even if another element of the // Make sure the struct picks up ompx_hold even if another element of the
// struct doesn't have ompx_hold. // struct doesn't have ompx_hold.
#pragma omp target map(tofrom : i) map(ompx_hold, tofrom : j) #pragma omp target map(tofrom : i) map(ompx_hold, tofrom : j)
Show All 9 Lines

clang/test/OpenMP/target_map_member_expr_array_section_codegen.cpp

// RUN: %clang_cc1 -verify -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s // RUN: %clang_cc1 -verify -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s
// RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s // RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
// expected-no-diagnostics // expected-no-diagnostics
#ifndef HEADER #ifndef HEADER
#define HEADER #define HEADER
// CHECK: [[SIZE_ENTER:@.+]] = private unnamed_addr global [2 x i64] [i64 0, i64 24]
// 0 = OMP_MAP_NONE // 0 = OMP_MAP_NONE
// 281474976710656 = 0x1000000000000 = OMP_MAP_MEMBER_OF of 1-st element // 281474976710656 = 0x1000000000000 = OMP_MAP_MEMBER_OF of 1-st element
// CHECK: [[MAP_ENTER:@.+]] = private unnamed_addr constant [2 x i64] [i64 0, i64 281474976710656] // CHECK: [[MAP_ENTER:@.+]] = private unnamed_addr constant [2 x i64] [i64 0, i64 281474976710656]
// CHECK: [[SIZE_EXIT:@.+]] = private unnamed_addr global [2 x i64] [i64 0, i64 24]
// 281474976710664 = 0x1000000000008 = OMP_MAP_MEMBER_OF of 1-st element | OMP_MAP_DELETE // 281474976710664 = 0x1000000000008 = OMP_MAP_MEMBER_OF of 1-st element | OMP_MAP_DELETE
// CHECK: [[MAP_EXIT:@.+]] = private unnamed_addr constant [2 x i64] [i64 0, i64 281474976710664] // CHECK: [[MAP_EXIT:@.+]] = private unnamed_addr constant [2 x i64] [i64 0, i64 281474976710664]
template <typename T> template <typename T>
struct S { struct S {
constexpr static int size = 6; constexpr static int size = 6;
T data[size]; T data[size];
}; };
template <typename T> template <typename T>
struct maptest { struct maptest {
S<T> s; S<T> s;
maptest() { maptest() {
// CHECK: [[BPTRS:%.+]] = alloca [2 x i8*], // CHECK: [[BPTRS:%.+]] = alloca [2 x i8*],
// CHECK: [[PTRS:%.+]] = alloca [2 x i8*], // CHECK: [[PTRS:%.+]] = alloca [2 x i8*],
// CHECK: [[SIZES:%.+]] = alloca [2 x i64],
// CHECK: getelementptr inbounds // CHECK: getelementptr inbounds
// CHECK: [[S_ADDR:%.+]] = getelementptr inbounds %struct.maptest, %struct.maptest* [[THIS:%.+]], i32 0, i32 0 // CHECK: [[S_ADDR:%.+]] = getelementptr inbounds %struct.maptest, %struct.maptest* [[THIS:%.+]], i32 0, i32 0
// CHECK: [[S_DATA_ADDR:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[S_ADDR]], i32 0, i32 0 // CHECK: [[S_DATA_ADDR:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[S_ADDR]], i32 0, i32 0
// CHECK: [[S_DATA_0_ADDR:%.+]] = getelementptr inbounds [6 x float], [6 x float]* [[S_DATA_ADDR]], i64 0, i64 0 // CHECK: [[S_DATA_0_ADDR:%.+]] = getelementptr inbounds [6 x float], [6 x float]* [[S_DATA_ADDR]], i64 0, i64 0
// SZ = &this->s.data[6]-&this->s.data[0] // SZ = &this->s.data[6]-&this->s.data[0]
// CHECK: [[S_ADDR:%.+]] = getelementptr inbounds %struct.maptest, %struct.maptest* [[THIS]], i32 0, i32 0 // CHECK: [[S_ADDR:%.+]] = getelementptr inbounds %struct.maptest, %struct.maptest* [[THIS]], i32 0, i32 0
// CHECK: [[S_DATA_ADDR:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[S_ADDR]], i32 0, i32 0 // CHECK: [[S_DATA_ADDR:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[S_ADDR]], i32 0, i32 0
// CHECK: [[S_DATA_5_ADDR:%.+]] = getelementptr inbounds [6 x float], [6 x float]* [[S_DATA_ADDR]], i64 0, i64 5 // CHECK: [[S_DATA_5_ADDR:%.+]] = getelementptr inbounds [6 x float], [6 x float]* [[S_DATA_ADDR]], i64 0, i64 5
// CHECK: [[S_DATA_6_ADDR:%.+]] = getelementptr float, float* [[S_DATA_5_ADDR]], i32 1 // CHECK: [[S_DATA_6_ADDR:%.+]] = getelementptr float, float* [[S_DATA_5_ADDR]], i32 1
// CHECK: [[BEG:%.+]] = bitcast float* [[S_DATA_0_ADDR]] to i8* // CHECK: [[BEG:%.+]] = bitcast float* [[S_DATA_0_ADDR]] to i8*
// CHECK: [[END:%.+]] = bitcast float* [[S_DATA_6_ADDR]] to i8* // CHECK: [[END:%.+]] = bitcast float* [[S_DATA_6_ADDR]] to i8*
// CHECK: [[END_BC:%.+]] = ptrtoint i8* [[END]] to i64 // CHECK: [[END_BC:%.+]] = ptrtoint i8* [[END]] to i64
// CHECK: [[BEG_BC:%.+]] = ptrtoint i8* [[BEG]] to i64 // CHECK: [[BEG_BC:%.+]] = ptrtoint i8* [[BEG]] to i64
// CHECK: [[DIFF:%.+]] = sub i64 [[END_BC]], [[BEG_BC]] // CHECK: [[DIFF:%.+]] = sub i64 [[END_BC]], [[BEG_BC]]
// CHECK: [[SZ:%.+]] = sdiv exact i64 [[DIFF]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK: [[SZ:%.+]] = sdiv exact i64 [[DIFF]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// Fill mapping arrays // Fill mapping arrays
// CHECK: [[BPTR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 0 // CHECK: [[BPTR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 0
// CHECK: [[BPTR0_THIS:%.+]] = bitcast i8** [[BPTR0]] to %struct.maptest** // CHECK: [[BPTR0_THIS:%.+]] = bitcast i8** [[BPTR0]] to %struct.maptest**
// CHECK: store %struct.maptest* [[THIS]], %struct.maptest** [[BPTR0_THIS]], // CHECK: store %struct.maptest* [[THIS]], %struct.maptest** [[BPTR0_THIS]],
// CHECK: [[PTR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 0 // CHECK: [[PTR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 0
// CHECK: [[PTR0_DATA:%.+]] = bitcast i8** [[PTR0]] to float** // CHECK: [[PTR0_DATA:%.+]] = bitcast i8** [[PTR0]] to float**
// CHECK: store float* [[S_DATA_0_ADDR]], float** [[PTR0_DATA]], // CHECK: store float* [[S_DATA_0_ADDR]], float** [[PTR0_DATA]],
// CHECK: [[SIZE0:%.+]] = getelementptr inbounds [2 x i64], [2 x i64]* [[SIZES]], i32 0, i32 0 // CHECK: store i64 [[SZ]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE_ENTER]], i32 0, i32 0),
// CHECK: store i64 [[SZ]], i64* [[SIZE0]],
// CHECK: [[BPTR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 1 // CHECK: [[BPTR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 1
// CHECK: [[BPTR1_THIS:%.+]] = bitcast i8** [[BPTR1]] to %struct.maptest** // CHECK: [[BPTR1_THIS:%.+]] = bitcast i8** [[BPTR1]] to %struct.maptest**
// CHECK: store %struct.maptest* [[THIS]], %struct.maptest** [[BPTR1_THIS]], // CHECK: store %struct.maptest* [[THIS]], %struct.maptest** [[BPTR1_THIS]],
// CHECK: [[PTR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 1 // CHECK: [[PTR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 1
// CHECK: [[PTR1_DATA:%.+]] = bitcast i8** [[PTR1]] to float** // CHECK: [[PTR1_DATA:%.+]] = bitcast i8** [[PTR1]] to float**
// CHECK: store float* [[S_DATA_0_ADDR]], float** [[PTR1_DATA]], // CHECK: store float* [[S_DATA_0_ADDR]], float** [[PTR1_DATA]],
// CHECK: [[SIZE1:%.+]] = getelementptr inbounds [2 x i64], [2 x i64]* [[SIZES]], i32 0, i32 1
// CHECK: store i64 24, i64* [[SIZE1]],
// CHECK: [[BPTR:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 0 // CHECK: [[BPTR:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 0
// CHECK: [[PTR:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 0 // CHECK: [[PTR:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 0
// CHECK: [[SIZE:%.+]] = getelementptr inbounds [2 x i64], [2 x i64]* [[SIZES]], i32 0, i32 0 // CHECK: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[BPTR]], i8** [[PTR]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE_ENTER]], i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[MAP_ENTER]], i32 0, i32 0), i8** null, i8** null)
// CHECK: call void @__tgt_target_data_begin_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[BPTR]], i8** [[PTR]], i64* [[SIZE]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[MAP_ENTER]], i32 0, i32 0), i8** null, i8** null)
#pragma omp target enter data map(alloc : s.data[:6]) #pragma omp target enter data map(alloc : s.data[:6])
} }
~maptest() { ~maptest() {
// CHECK: [[BPTRS:%.+]] = alloca [2 x i8*], // CHECK: [[BPTRS:%.+]] = alloca [2 x i8*],
// CHECK: [[PTRS:%.+]] = alloca [2 x i8*], // CHECK: [[PTRS:%.+]] = alloca [2 x i8*],
// CHECK: [[SIZE:%.+]] = alloca [2 x i64],
// CHECK: [[S_ADDR:%.+]] = getelementptr inbounds %struct.maptest, %struct.maptest* [[THIS:%.+]], i32 0, i32 0 // CHECK: [[S_ADDR:%.+]] = getelementptr inbounds %struct.maptest, %struct.maptest* [[THIS:%.+]], i32 0, i32 0
// CHECK: [[S_DATA_ADDR:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[S_ADDR]], i32 0, i32 0 // CHECK: [[S_DATA_ADDR:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[S_ADDR]], i32 0, i32 0
// CHECK: [[S_DATA_0_ADDR:%.+]] = getelementptr inbounds [6 x float], [6 x float]* [[S_DATA_ADDR]], i64 0, i64 0 // CHECK: [[S_DATA_0_ADDR:%.+]] = getelementptr inbounds [6 x float], [6 x float]* [[S_DATA_ADDR]], i64 0, i64 0
// SZ = &this->s.data[6]-&this->s.data[0] // SZ = &this->s.data[6]-&this->s.data[0]
// CHECK: [[S_ADDR:%.+]] = getelementptr inbounds %struct.maptest, %struct.maptest* [[THIS]], i32 0, i32 0 // CHECK: [[S_ADDR:%.+]] = getelementptr inbounds %struct.maptest, %struct.maptest* [[THIS]], i32 0, i32 0
// CHECK: [[S_DATA_ADDR:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[S_ADDR]], i32 0, i32 0 // CHECK: [[S_DATA_ADDR:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[S_ADDR]], i32 0, i32 0
// CHECK: [[S_DATA_5_ADDR:%.+]] = getelementptr inbounds [6 x float], [6 x float]* [[S_DATA_ADDR]], i64 0, i64 5 // CHECK: [[S_DATA_5_ADDR:%.+]] = getelementptr inbounds [6 x float], [6 x float]* [[S_DATA_ADDR]], i64 0, i64 5
// CHECK: [[S_DATA_6_ADDR:%.+]] = getelementptr float, float* [[S_DATA_5_ADDR]], i32 1 // CHECK: [[S_DATA_6_ADDR:%.+]] = getelementptr float, float* [[S_DATA_5_ADDR]], i32 1
// CHECK: [[BEG:%.+]] = bitcast float* [[S_DATA_0_ADDR]] to i8* // CHECK: [[BEG:%.+]] = bitcast float* [[S_DATA_0_ADDR]] to i8*
// CHECK: [[END:%.+]] = bitcast float* [[S_DATA_6_ADDR]] to i8* // CHECK: [[END:%.+]] = bitcast float* [[S_DATA_6_ADDR]] to i8*
// CHECK: [[END_BC:%.+]] = ptrtoint i8* [[END]] to i64 // CHECK: [[END_BC:%.+]] = ptrtoint i8* [[END]] to i64
// CHECK: [[BEG_BC:%.+]] = ptrtoint i8* [[BEG]] to i64 // CHECK: [[BEG_BC:%.+]] = ptrtoint i8* [[BEG]] to i64
// CHECK: [[DIFF:%.+]] = sub i64 [[END_BC]], [[BEG_BC]] // CHECK: [[DIFF:%.+]] = sub i64 [[END_BC]], [[BEG_BC]]
// CHECK: [[SZ:%.+]] = sdiv exact i64 [[DIFF]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK: [[SZ:%.+]] = sdiv exact i64 [[DIFF]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// Fill mapping arrays // Fill mapping arrays
// CHECK: [[BPTR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 0 // CHECK: [[BPTR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 0
// CHECK: [[BPTR0_THIS:%.+]] = bitcast i8** [[BPTR0]] to %struct.maptest** // CHECK: [[BPTR0_THIS:%.+]] = bitcast i8** [[BPTR0]] to %struct.maptest**
// CHECK: store %struct.maptest* [[THIS]], %struct.maptest** [[BPTR0_THIS]], // CHECK: store %struct.maptest* [[THIS]], %struct.maptest** [[BPTR0_THIS]],
// CHECK: [[PTR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 0 // CHECK: [[PTR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 0
// CHECK: [[PTR0_DATA:%.+]] = bitcast i8** [[PTR0]] to float** // CHECK: [[PTR0_DATA:%.+]] = bitcast i8** [[PTR0]] to float**
// CHECK: store float* [[S_DATA_0_ADDR]], float** [[PTR0_DATA]], // CHECK: store float* [[S_DATA_0_ADDR]], float** [[PTR0_DATA]],
// CHECK: [[SIZE0:%.+]] = getelementptr inbounds [2 x i64], [2 x i64]* [[SIZES]], i32 0, i32 0 // CHECK: store i64 [[SZ]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE_EXIT]], i32 0, i32 0),
// CHECK: store i64 [[SZ]], i64* [[SIZE0]],
// CHECK: [[BPTR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 1 // CHECK: [[BPTR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 1
// CHECK: [[BPTR1_THIS:%.+]] = bitcast i8** [[BPTR1]] to %struct.maptest** // CHECK: [[BPTR1_THIS:%.+]] = bitcast i8** [[BPTR1]] to %struct.maptest**
// CHECK: store %struct.maptest* [[THIS]], %struct.maptest** [[BPTR1_THIS]], // CHECK: store %struct.maptest* [[THIS]], %struct.maptest** [[BPTR1_THIS]],
// CHECK: [[PTR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 1 // CHECK: [[PTR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 1
// CHECK: [[PTR1_DATA:%.+]] = bitcast i8** [[PTR1]] to float** // CHECK: [[PTR1_DATA:%.+]] = bitcast i8** [[PTR1]] to float**
// CHECK: store float* [[S_DATA_0_ADDR]], float** [[PTR1_DATA]], // CHECK: store float* [[S_DATA_0_ADDR]], float** [[PTR1_DATA]],
// CHECK: [[SIZE1:%.+]] = getelementptr inbounds [2 x i64], [2 x i64]* [[SIZES]], i32 0, i32 1
// CHECK: store i64 24, i64* [[SIZE1]],
// CHECK: [[BPTR:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 0 // CHECK: [[BPTR:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BPTRS]], i32 0, i32 0
// CHECK: [[PTR:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 0 // CHECK: [[PTR:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS]], i32 0, i32 0
// CHECK: [[SIZE:%.+]] = getelementptr inbounds [2 x i64], [2 x i64]* [[SIZES]], i32 0, i32 0 // CHECK: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[BPTR]], i8** [[PTR]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE_EXIT]], i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[MAP_EXIT]], i32 0, i32 0), i8** null, i8** null)
// CHECK: call void @__tgt_target_data_end_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[BPTR]], i8** [[PTR]], i64* [[SIZE]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[MAP_EXIT]], i32 0, i32 0), i8** null, i8** null)
#pragma omp target exit data map(delete : s.data[:6]) #pragma omp target exit data map(delete : s.data[:6])
} }
}; };
maptest<float> a; maptest<float> a;
#endif #endif

clang/test/OpenMP/target_parallel_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 312 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[AA_CASTED5:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_CASTED5:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS7:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS7:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS8:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS8:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS9:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS9:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[A_CASTED12:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_CASTED12:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [9 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [9 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [9 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [9 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [9 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [9 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]]) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
// CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 0, i32* [[A]], align 4 // CHECK1-NEXT: store i32 0, i32* [[A]], align 4
// CHECK1-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK1-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK1-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK1-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP47:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK1-NEXT: [[TMP47:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK1-NEXT: [[TMP48:%.*]] = mul nuw i64 [[TMP47]], 8 // CHECK1-NEXT: [[TMP48:%.*]] = mul nuw i64 [[TMP47]], 8
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK1-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i64* // CHECK1-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i64*
// CHECK1-NEXT: store i64 [[TMP44]], i64* [[TMP50]], align 8 // CHECK1-NEXT: store i64 [[TMP44]], i64* [[TMP50]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0 // CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK1-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i64* // CHECK1-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i64*
// CHECK1-NEXT: store i64 [[TMP44]], i64* [[TMP52]], align 8 // CHECK1-NEXT: store i64 [[TMP44]], i64* [[TMP52]], align 8
// CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 0
// CHECK1-NEXT: store i64 4, i64* [[TMP53]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP53]], align 8
// CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 0 // CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 1
// CHECK1-NEXT: store i8* null, i8** [[TMP54]], align 8 // CHECK1-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to [10 x float]**
// CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 1 // CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP55]], align 8
// CHECK1-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to [10 x float]** // CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 1
// CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP56]], align 8 // CHECK1-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to [10 x float]**
// CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 1 // CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP57]], align 8
// CHECK1-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to [10 x float]** // CHECK1-NEXT: [[TMP58:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 1
// CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP58]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP58]], align 8
// CHECK1-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK1-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 2
// CHECK1-NEXT: store i64 40, i64* [[TMP59]], align 8 // CHECK1-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i64*
// CHECK1-NEXT: [[TMP60:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 1 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP60]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP60]], align 8 // CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 2
// CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 2
// CHECK1-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64* // CHECK1-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP62]], align 8 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP62]], align 8
// CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 2 // CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 2
// CHECK1-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64* // CHECK1-NEXT: store i8* null, i8** [[TMP63]], align 8
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP64]], align 8 // CHECK1-NEXT: [[TMP64:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 3
// CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK1-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to float**
// CHECK1-NEXT: store i64 8, i64* [[TMP65]], align 8 // CHECK1-NEXT: store float* [[VLA]], float** [[TMP65]], align 8
// CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 2 // CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 3
// CHECK1-NEXT: store i8* null, i8** [[TMP66]], align 8 // CHECK1-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to float**
// CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 3 // CHECK1-NEXT: store float* [[VLA]], float** [[TMP67]], align 8
// CHECK1-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to float** // CHECK1-NEXT: store i64 [[TMP46]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK1-NEXT: store float* [[VLA]], float** [[TMP68]], align 8 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 3
// CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 3 // CHECK1-NEXT: store i8* null, i8** [[TMP68]], align 8
// CHECK1-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to float** // CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 4
// CHECK1-NEXT: store float* [[VLA]], float** [[TMP70]], align 8 // CHECK1-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to [5 x [10 x double]]**
// CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP70]], align 8
// CHECK1-NEXT: store i64 [[TMP46]], i64* [[TMP71]], align 8 // CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 4
// CHECK1-NEXT: [[TMP72:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 3 // CHECK1-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [5 x [10 x double]]**
// CHECK1-NEXT: store i8* null, i8** [[TMP72]], align 8 // CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP72]], align 8
// CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 4 // CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 4
// CHECK1-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [5 x [10 x double]]** // CHECK1-NEXT: store i8* null, i8** [[TMP73]], align 8
// CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP74]], align 8 // CHECK1-NEXT: [[TMP74:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 5
// CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 4 // CHECK1-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64*
// CHECK1-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [5 x [10 x double]]** // CHECK1-NEXT: store i64 5, i64* [[TMP75]], align 8
// CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP76]], align 8 // CHECK1-NEXT: [[TMP76:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 5
// CHECK1-NEXT: [[TMP77:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK1-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK1-NEXT: store i64 400, i64* [[TMP77]], align 8 // CHECK1-NEXT: store i64 5, i64* [[TMP77]], align 8
// CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 4 // CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 5
// CHECK1-NEXT: store i8* null, i8** [[TMP78]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP78]], align 8
// CHECK1-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 5 // CHECK1-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 6
// CHECK1-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i64* // CHECK1-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i64*
// CHECK1-NEXT: store i64 5, i64* [[TMP80]], align 8 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP80]], align 8
// CHECK1-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 5 // CHECK1-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 6
// CHECK1-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i64* // CHECK1-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i64*
// CHECK1-NEXT: store i64 5, i64* [[TMP82]], align 8 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP82]], align 8
// CHECK1-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK1-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 6
// CHECK1-NEXT: store i64 8, i64* [[TMP83]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP83]], align 8
// CHECK1-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 5 // CHECK1-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 7
// CHECK1-NEXT: store i8* null, i8** [[TMP84]], align 8 // CHECK1-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to double**
// CHECK1-NEXT: [[TMP85:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 6 // CHECK1-NEXT: store double* [[VLA1]], double** [[TMP85]], align 8
// CHECK1-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i64* // CHECK1-NEXT: [[TMP86:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 7
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP86]], align 8 // CHECK1-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to double**
// CHECK1-NEXT: [[TMP87:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 6 // CHECK1-NEXT: store double* [[VLA1]], double** [[TMP87]], align 8
// CHECK1-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK1-NEXT: store i64 [[TMP48]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP88]], align 8 // CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 7
// CHECK1-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK1-NEXT: store i8* null, i8** [[TMP88]], align 8
// CHECK1-NEXT: store i64 8, i64* [[TMP89]], align 8 // CHECK1-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 8
// CHECK1-NEXT: [[TMP90:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 6 // CHECK1-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to %struct.TT**
// CHECK1-NEXT: store i8* null, i8** [[TMP90]], align 8 // CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP90]], align 8
// CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 7 // CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 8
// CHECK1-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to double** // CHECK1-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to %struct.TT**
// CHECK1-NEXT: store double* [[VLA1]], double** [[TMP92]], align 8 // CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP92]], align 8
// CHECK1-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 7 // CHECK1-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 8
// CHECK1-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to double** // CHECK1-NEXT: store i8* null, i8** [[TMP93]], align 8
// CHECK1-NEXT: store double* [[VLA1]], double** [[TMP94]], align 8 // CHECK1-NEXT: [[TMP94:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK1-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK1-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP48]], i64* [[TMP95]], align 8 // CHECK1-NEXT: [[TMP96:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP94]], i8** [[TMP95]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP96:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 7 // CHECK1-NEXT: [[TMP97:%.*]] = icmp ne i32 [[TMP96]], 0
// CHECK1-NEXT: store i8* null, i8** [[TMP96]], align 8 // CHECK1-NEXT: br i1 [[TMP97]], label [[OMP_OFFLOAD_FAILED19:%.*]], label [[OMP_OFFLOAD_CONT20:%.*]]
// CHECK1-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 8
// CHECK1-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to %struct.TT**
// CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP98]], align 8
// CHECK1-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 8
// CHECK1-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to %struct.TT**
// CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP100]], align 8
// CHECK1-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK1-NEXT: store i64 16, i64* [[TMP101]], align 8
// CHECK1-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 8
// CHECK1-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK1-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK1-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK1-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK1-NEXT: [[TMP106:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP103]], i8** [[TMP104]], i64* [[TMP105]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP107:%.*]] = icmp ne i32 [[TMP106]], 0
// CHECK1-NEXT: br i1 [[TMP107]], label [[OMP_OFFLOAD_FAILED19:%.*]], label [[OMP_OFFLOAD_CONT20:%.*]]
// CHECK1: omp_offload.failed19: // CHECK1: omp_offload.failed19:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT20]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT20]]
// CHECK1: omp_offload.cont20: // CHECK1: omp_offload.cont20:
// CHECK1-NEXT: br label [[OMP_IF_END22:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END22:%.*]]
// CHECK1: omp_if.else21: // CHECK1: omp_if.else21:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_IF_END22]] // CHECK1-NEXT: br label [[OMP_IF_END22]]
// CHECK1: omp_if.end22: // CHECK1: omp_if.end22:
// CHECK1-NEXT: [[TMP108:%.*]] = load i32, i32* [[A]], align 4 // CHECK1-NEXT: [[TMP98:%.*]] = load i32, i32* [[A]], align 4
// CHECK1-NEXT: [[TMP109:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP99:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP109]]) // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP99]])
// CHECK1-NEXT: ret i32 [[TMP108]] // CHECK1-NEXT: ret i32 [[TMP98]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100
// CHECK1-SAME: () #[[ATTR2:[0-9]+]] { // CHECK1-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
▲ Show 20 Lines • Show All 321 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK1-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK1-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK1-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK1-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 14 Lines
// CHECK1-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK1-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK1-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK1-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK1-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK1-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK1-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK1-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK1-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK1-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK1-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK1-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK1-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK1-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK1-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK1-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK1-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK1-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK1-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK1-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK1-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK1-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8 // CHECK1-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK1-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK1-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK1-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_IF_END]] // CHECK1-NEXT: br label [[OMP_IF_END]]
// CHECK1: omp_if.end: // CHECK1: omp_if.end:
// CHECK1-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK1-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK1-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK1-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK1-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK1-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK1-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK1-NEXT: ret i32 [[ADD4]] // CHECK1-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK1-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK1-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK1-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK1-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK1-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
Show All 22 Lines
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK1-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK1-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK1-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK1-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK1-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK1-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK1-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[AA_CASTED5:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_CASTED5:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS7:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS7:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS8:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS8:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS9:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS9:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[A_CASTED12:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_CASTED12:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [9 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [9 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [9 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [9 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [9 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [9 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]]) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
// CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 0, i32* [[A]], align 4 // CHECK2-NEXT: store i32 0, i32* [[A]], align 4
// CHECK2-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK2-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK2-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK2-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP47:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK2-NEXT: [[TMP47:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK2-NEXT: [[TMP48:%.*]] = mul nuw i64 [[TMP47]], 8 // CHECK2-NEXT: [[TMP48:%.*]] = mul nuw i64 [[TMP47]], 8
// CHECK2-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0 // CHECK2-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK2-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i64* // CHECK2-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i64*
// CHECK2-NEXT: store i64 [[TMP44]], i64* [[TMP50]], align 8 // CHECK2-NEXT: store i64 [[TMP44]], i64* [[TMP50]], align 8
// CHECK2-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0 // CHECK2-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK2-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i64* // CHECK2-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i64*
// CHECK2-NEXT: store i64 [[TMP44]], i64* [[TMP52]], align 8 // CHECK2-NEXT: store i64 [[TMP44]], i64* [[TMP52]], align 8
// CHECK2-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK2-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 0
// CHECK2-NEXT: store i64 4, i64* [[TMP53]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP53]], align 8
// CHECK2-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 0 // CHECK2-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP54]], align 8 // CHECK2-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to [10 x float]**
// CHECK2-NEXT: [[TMP55:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 1 // CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP55]], align 8
// CHECK2-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to [10 x float]** // CHECK2-NEXT: [[TMP56:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 1
// CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP56]], align 8 // CHECK2-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to [10 x float]**
// CHECK2-NEXT: [[TMP57:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 1 // CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP57]], align 8
// CHECK2-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to [10 x float]** // CHECK2-NEXT: [[TMP58:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 1
// CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP58]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP58]], align 8
// CHECK2-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK2-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 2
// CHECK2-NEXT: store i64 40, i64* [[TMP59]], align 8 // CHECK2-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i64*
// CHECK2-NEXT: [[TMP60:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 1 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP60]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP60]], align 8 // CHECK2-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 2
// CHECK2-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 2
// CHECK2-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64* // CHECK2-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP62]], align 8 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP62]], align 8
// CHECK2-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 2 // CHECK2-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 2
// CHECK2-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64* // CHECK2-NEXT: store i8* null, i8** [[TMP63]], align 8
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP64]], align 8 // CHECK2-NEXT: [[TMP64:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 3
// CHECK2-NEXT: [[TMP65:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK2-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to float**
// CHECK2-NEXT: store i64 8, i64* [[TMP65]], align 8 // CHECK2-NEXT: store float* [[VLA]], float** [[TMP65]], align 8
// CHECK2-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 2 // CHECK2-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 3
// CHECK2-NEXT: store i8* null, i8** [[TMP66]], align 8 // CHECK2-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to float**
// CHECK2-NEXT: [[TMP67:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 3 // CHECK2-NEXT: store float* [[VLA]], float** [[TMP67]], align 8
// CHECK2-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to float** // CHECK2-NEXT: store i64 [[TMP46]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK2-NEXT: store float* [[VLA]], float** [[TMP68]], align 8 // CHECK2-NEXT: [[TMP68:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 3
// CHECK2-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 3 // CHECK2-NEXT: store i8* null, i8** [[TMP68]], align 8
// CHECK2-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to float** // CHECK2-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 4
// CHECK2-NEXT: store float* [[VLA]], float** [[TMP70]], align 8 // CHECK2-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to [5 x [10 x double]]**
// CHECK2-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP70]], align 8
// CHECK2-NEXT: store i64 [[TMP46]], i64* [[TMP71]], align 8 // CHECK2-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 4
// CHECK2-NEXT: [[TMP72:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 3 // CHECK2-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [5 x [10 x double]]**
// CHECK2-NEXT: store i8* null, i8** [[TMP72]], align 8 // CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP72]], align 8
// CHECK2-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 4 // CHECK2-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 4
// CHECK2-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [5 x [10 x double]]** // CHECK2-NEXT: store i8* null, i8** [[TMP73]], align 8
// CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP74]], align 8 // CHECK2-NEXT: [[TMP74:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 5
// CHECK2-NEXT: [[TMP75:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 4 // CHECK2-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64*
// CHECK2-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [5 x [10 x double]]** // CHECK2-NEXT: store i64 5, i64* [[TMP75]], align 8
// CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP76]], align 8 // CHECK2-NEXT: [[TMP76:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 5
// CHECK2-NEXT: [[TMP77:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK2-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK2-NEXT: store i64 400, i64* [[TMP77]], align 8 // CHECK2-NEXT: store i64 5, i64* [[TMP77]], align 8
// CHECK2-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 4 // CHECK2-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 5
// CHECK2-NEXT: store i8* null, i8** [[TMP78]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP78]], align 8
// CHECK2-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 5 // CHECK2-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 6
// CHECK2-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i64* // CHECK2-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i64*
// CHECK2-NEXT: store i64 5, i64* [[TMP80]], align 8 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP80]], align 8
// CHECK2-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 5 // CHECK2-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 6
// CHECK2-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i64* // CHECK2-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i64*
// CHECK2-NEXT: store i64 5, i64* [[TMP82]], align 8 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP82]], align 8
// CHECK2-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK2-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 6
// CHECK2-NEXT: store i64 8, i64* [[TMP83]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP83]], align 8
// CHECK2-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 5 // CHECK2-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 7
// CHECK2-NEXT: store i8* null, i8** [[TMP84]], align 8 // CHECK2-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to double**
// CHECK2-NEXT: [[TMP85:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 6 // CHECK2-NEXT: store double* [[VLA1]], double** [[TMP85]], align 8
// CHECK2-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i64* // CHECK2-NEXT: [[TMP86:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 7
// CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP86]], align 8 // CHECK2-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to double**
// CHECK2-NEXT: [[TMP87:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 6 // CHECK2-NEXT: store double* [[VLA1]], double** [[TMP87]], align 8
// CHECK2-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK2-NEXT: store i64 [[TMP48]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP88]], align 8 // CHECK2-NEXT: [[TMP88:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 7
// CHECK2-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK2-NEXT: store i8* null, i8** [[TMP88]], align 8
// CHECK2-NEXT: store i64 8, i64* [[TMP89]], align 8 // CHECK2-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 8
// CHECK2-NEXT: [[TMP90:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 6 // CHECK2-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to %struct.TT**
// CHECK2-NEXT: store i8* null, i8** [[TMP90]], align 8 // CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP90]], align 8
// CHECK2-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 7 // CHECK2-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 8
// CHECK2-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to double** // CHECK2-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to %struct.TT**
// CHECK2-NEXT: store double* [[VLA1]], double** [[TMP92]], align 8 // CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP92]], align 8
// CHECK2-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 7 // CHECK2-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 8
// CHECK2-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to double** // CHECK2-NEXT: store i8* null, i8** [[TMP93]], align 8
// CHECK2-NEXT: store double* [[VLA1]], double** [[TMP94]], align 8 // CHECK2-NEXT: [[TMP94:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK2-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK2-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK2-NEXT: store i64 [[TMP48]], i64* [[TMP95]], align 8 // CHECK2-NEXT: [[TMP96:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP94]], i8** [[TMP95]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP96:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 7 // CHECK2-NEXT: [[TMP97:%.*]] = icmp ne i32 [[TMP96]], 0
// CHECK2-NEXT: store i8* null, i8** [[TMP96]], align 8 // CHECK2-NEXT: br i1 [[TMP97]], label [[OMP_OFFLOAD_FAILED19:%.*]], label [[OMP_OFFLOAD_CONT20:%.*]]
// CHECK2-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 8
// CHECK2-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to %struct.TT**
// CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP98]], align 8
// CHECK2-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 8
// CHECK2-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to %struct.TT**
// CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP100]], align 8
// CHECK2-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK2-NEXT: store i64 16, i64* [[TMP101]], align 8
// CHECK2-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 8
// CHECK2-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK2-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK2-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK2-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK2-NEXT: [[TMP106:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP103]], i8** [[TMP104]], i64* [[TMP105]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP107:%.*]] = icmp ne i32 [[TMP106]], 0
// CHECK2-NEXT: br i1 [[TMP107]], label [[OMP_OFFLOAD_FAILED19:%.*]], label [[OMP_OFFLOAD_CONT20:%.*]]
// CHECK2: omp_offload.failed19: // CHECK2: omp_offload.failed19:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT20]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT20]]
// CHECK2: omp_offload.cont20: // CHECK2: omp_offload.cont20:
// CHECK2-NEXT: br label [[OMP_IF_END22:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END22:%.*]]
// CHECK2: omp_if.else21: // CHECK2: omp_if.else21:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_IF_END22]] // CHECK2-NEXT: br label [[OMP_IF_END22]]
// CHECK2: omp_if.end22: // CHECK2: omp_if.end22:
// CHECK2-NEXT: [[TMP108:%.*]] = load i32, i32* [[A]], align 4 // CHECK2-NEXT: [[TMP98:%.*]] = load i32, i32* [[A]], align 4
// CHECK2-NEXT: [[TMP109:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: [[TMP99:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP109]]) // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP99]])
// CHECK2-NEXT: ret i32 [[TMP108]] // CHECK2-NEXT: ret i32 [[TMP98]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100
// CHECK2-SAME: () #[[ATTR2:[0-9]+]] { // CHECK2-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
▲ Show 20 Lines • Show All 321 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK2-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK2-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK2-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK2-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 14 Lines
// CHECK2-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK2-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK2-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK2-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK2-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK2-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK2-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK2-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK2-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK2-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK2-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK2-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK2-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK2-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK2-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK2-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK2-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK2-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK2-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK2-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK2-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK2-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK2-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK2-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK2-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK2-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK2-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK2-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK2-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK2-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8 // CHECK2-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK2-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK2-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK2-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK2-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK2-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK2-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK2-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_IF_END]] // CHECK2-NEXT: br label [[OMP_IF_END]]
// CHECK2: omp_if.end: // CHECK2: omp_if.end:
// CHECK2-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK2-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK2-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK2-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK2-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK2-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK2-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK2-NEXT: ret i32 [[ADD4]] // CHECK2-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK2-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK2-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK2-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK2-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK2-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK2-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK2-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK2-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK2-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
Show All 22 Lines
// CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK2-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK2-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK2-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK2-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK2-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK2-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK2-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[AA_CASTED3:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_CASTED3:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[A_CASTED10:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_CASTED10:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS13:%.*]] = alloca [9 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS13:%.*]] = alloca [9 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS14:%.*]] = alloca [9 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS14:%.*]] = alloca [9 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS15:%.*]] = alloca [9 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS15:%.*]] = alloca [9 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]]) // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
// CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[A]], align 4 // CHECK3-NEXT: store i32 0, i32* [[A]], align 4
// CHECK3-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK3-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK3-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK3-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 85 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP47:%.*]] = mul nuw i32 [[TMP46]], 8 // CHECK3-NEXT: [[TMP47:%.*]] = mul nuw i32 [[TMP46]], 8
// CHECK3-NEXT: [[TMP48:%.*]] = sext i32 [[TMP47]] to i64 // CHECK3-NEXT: [[TMP48:%.*]] = sext i32 [[TMP47]] to i64
// CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0 // CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK3-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32* // CHECK3-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32*
// CHECK3-NEXT: store i32 [[TMP42]], i32* [[TMP50]], align 4 // CHECK3-NEXT: store i32 [[TMP42]], i32* [[TMP50]], align 4
// CHECK3-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0 // CHECK3-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK3-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32* // CHECK3-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32*
// CHECK3-NEXT: store i32 [[TMP42]], i32* [[TMP52]], align 4 // CHECK3-NEXT: store i32 [[TMP42]], i32* [[TMP52]], align 4
// CHECK3-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK3-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 0
// CHECK3-NEXT: store i64 4, i64* [[TMP53]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP53]], align 4
// CHECK3-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 0 // CHECK3-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK3-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to [10 x float]**
// CHECK3-NEXT: [[TMP55:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 1 // CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP55]], align 4
// CHECK3-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to [10 x float]** // CHECK3-NEXT: [[TMP56:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 1
// CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP56]], align 4 // CHECK3-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to [10 x float]**
// CHECK3-NEXT: [[TMP57:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 1 // CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP57]], align 4
// CHECK3-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to [10 x float]** // CHECK3-NEXT: [[TMP58:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 1
// CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP58]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP58]], align 4
// CHECK3-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK3-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 2
// CHECK3-NEXT: store i64 40, i64* [[TMP59]], align 4 // CHECK3-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i32*
// CHECK3-NEXT: [[TMP60:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP60]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP60]], align 4 // CHECK3-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 2
// CHECK3-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 2
// CHECK3-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i32* // CHECK3-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP62]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP62]], align 4
// CHECK3-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 2 // CHECK3-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 2
// CHECK3-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32* // CHECK3-NEXT: store i8* null, i8** [[TMP63]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP64]], align 4 // CHECK3-NEXT: [[TMP64:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 3
// CHECK3-NEXT: [[TMP65:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK3-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to float**
// CHECK3-NEXT: store i64 4, i64* [[TMP65]], align 4 // CHECK3-NEXT: store float* [[VLA]], float** [[TMP65]], align 4
// CHECK3-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 2 // CHECK3-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP66]], align 4 // CHECK3-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to float**
// CHECK3-NEXT: [[TMP67:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 3 // CHECK3-NEXT: store float* [[VLA]], float** [[TMP67]], align 4
// CHECK3-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to float** // CHECK3-NEXT: store i64 [[TMP45]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK3-NEXT: store float* [[VLA]], float** [[TMP68]], align 4 // CHECK3-NEXT: [[TMP68:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 3
// CHECK3-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 3 // CHECK3-NEXT: store i8* null, i8** [[TMP68]], align 4
// CHECK3-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to float** // CHECK3-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 4
// CHECK3-NEXT: store float* [[VLA]], float** [[TMP70]], align 4 // CHECK3-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to [5 x [10 x double]]**
// CHECK3-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP70]], align 4
// CHECK3-NEXT: store i64 [[TMP45]], i64* [[TMP71]], align 4 // CHECK3-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 4
// CHECK3-NEXT: [[TMP72:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 3 // CHECK3-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [5 x [10 x double]]**
// CHECK3-NEXT: store i8* null, i8** [[TMP72]], align 4 // CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP72]], align 4
// CHECK3-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 4 // CHECK3-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 4
// CHECK3-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [5 x [10 x double]]** // CHECK3-NEXT: store i8* null, i8** [[TMP73]], align 4
// CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP74]], align 4 // CHECK3-NEXT: [[TMP74:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 5
// CHECK3-NEXT: [[TMP75:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 4 // CHECK3-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32*
// CHECK3-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [5 x [10 x double]]** // CHECK3-NEXT: store i32 5, i32* [[TMP75]], align 4
// CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP76]], align 4 // CHECK3-NEXT: [[TMP76:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 5
// CHECK3-NEXT: [[TMP77:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK3-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32*
// CHECK3-NEXT: store i64 400, i64* [[TMP77]], align 4 // CHECK3-NEXT: store i32 5, i32* [[TMP77]], align 4
// CHECK3-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 4 // CHECK3-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 5
// CHECK3-NEXT: store i8* null, i8** [[TMP78]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP78]], align 4
// CHECK3-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 5 // CHECK3-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 6
// CHECK3-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK3-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32*
// CHECK3-NEXT: store i32 5, i32* [[TMP80]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP80]], align 4
// CHECK3-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 5 // CHECK3-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 6
// CHECK3-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32* // CHECK3-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32*
// CHECK3-NEXT: store i32 5, i32* [[TMP82]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP82]], align 4
// CHECK3-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK3-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 6
// CHECK3-NEXT: store i64 4, i64* [[TMP83]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP83]], align 4
// CHECK3-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 5 // CHECK3-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 7
// CHECK3-NEXT: store i8* null, i8** [[TMP84]], align 4 // CHECK3-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to double**
// CHECK3-NEXT: [[TMP85:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 6 // CHECK3-NEXT: store double* [[VLA1]], double** [[TMP85]], align 4
// CHECK3-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32* // CHECK3-NEXT: [[TMP86:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 7
// CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP86]], align 4 // CHECK3-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to double**
// CHECK3-NEXT: [[TMP87:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 6 // CHECK3-NEXT: store double* [[VLA1]], double** [[TMP87]], align 4
// CHECK3-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32* // CHECK3-NEXT: store i64 [[TMP48]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP88]], align 4 // CHECK3-NEXT: [[TMP88:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 7
// CHECK3-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK3-NEXT: store i8* null, i8** [[TMP88]], align 4
// CHECK3-NEXT: store i64 4, i64* [[TMP89]], align 4 // CHECK3-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 8
// CHECK3-NEXT: [[TMP90:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 6 // CHECK3-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to %struct.TT**
// CHECK3-NEXT: store i8* null, i8** [[TMP90]], align 4 // CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP90]], align 4
// CHECK3-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 7 // CHECK3-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 8
// CHECK3-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to double** // CHECK3-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to %struct.TT**
// CHECK3-NEXT: store double* [[VLA1]], double** [[TMP92]], align 4 // CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP92]], align 4
// CHECK3-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 7 // CHECK3-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 8
// CHECK3-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to double** // CHECK3-NEXT: store i8* null, i8** [[TMP93]], align 4
// CHECK3-NEXT: store double* [[VLA1]], double** [[TMP94]], align 4 // CHECK3-NEXT: [[TMP94:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK3-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK3-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK3-NEXT: store i64 [[TMP48]], i64* [[TMP95]], align 4 // CHECK3-NEXT: [[TMP96:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP94]], i8** [[TMP95]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP96:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 7 // CHECK3-NEXT: [[TMP97:%.*]] = icmp ne i32 [[TMP96]], 0
// CHECK3-NEXT: store i8* null, i8** [[TMP96]], align 4 // CHECK3-NEXT: br i1 [[TMP97]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK3-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 8
// CHECK3-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to %struct.TT**
// CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP98]], align 4
// CHECK3-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 8
// CHECK3-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to %struct.TT**
// CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP100]], align 4
// CHECK3-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK3-NEXT: store i64 12, i64* [[TMP101]], align 4
// CHECK3-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 8
// CHECK3-NEXT: store i8* null, i8** [[TMP102]], align 4
// CHECK3-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK3-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK3-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK3-NEXT: [[TMP106:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP103]], i8** [[TMP104]], i64* [[TMP105]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP107:%.*]] = icmp ne i32 [[TMP106]], 0
// CHECK3-NEXT: br i1 [[TMP107]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK3: omp_offload.failed16: // CHECK3: omp_offload.failed16:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT17]]
// CHECK3: omp_offload.cont17: // CHECK3: omp_offload.cont17:
// CHECK3-NEXT: br label [[OMP_IF_END19:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END19:%.*]]
// CHECK3: omp_if.else18: // CHECK3: omp_if.else18:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_IF_END19]] // CHECK3-NEXT: br label [[OMP_IF_END19]]
// CHECK3: omp_if.end19: // CHECK3: omp_if.end19:
// CHECK3-NEXT: [[TMP108:%.*]] = load i32, i32* [[A]], align 4 // CHECK3-NEXT: [[TMP98:%.*]] = load i32, i32* [[A]], align 4
// CHECK3-NEXT: [[TMP109:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: [[TMP99:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP109]]) // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP99]])
// CHECK3-NEXT: ret i32 [[TMP108]] // CHECK3-NEXT: ret i32 [[TMP98]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100
// CHECK3-SAME: () #[[ATTR2:[0-9]+]] { // CHECK3-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
▲ Show 20 Lines • Show All 312 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK3-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK3-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK3-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 13 Lines
// CHECK3-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK3-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK3-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK3-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK3-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK3-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK3-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK3-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK3-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK3-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK3-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK3-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK3-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK3-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK3-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK3-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK3-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK3-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK3-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK3-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK3-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK3-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK3-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4 // CHECK3-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK3-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK3-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK3-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK3-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_IF_END]] // CHECK3-NEXT: br label [[OMP_IF_END]]
// CHECK3: omp_if.end: // CHECK3: omp_if.end:
// CHECK3-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK3-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK3-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK3-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK3-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK3-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK3-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK3-NEXT: ret i32 [[ADD3]] // CHECK3-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK3-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK3-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK3-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK3-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK3-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
Show All 20 Lines
// CHECK3-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK3-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 49 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK3-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK3-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK3-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK3-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK3-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK3-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK3-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
▲ Show 20 Lines • Show All 41 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK3-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK3-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
▲ Show 20 Lines • Show All 47 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[AA_CASTED3:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_CASTED3:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[A_CASTED10:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_CASTED10:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS13:%.*]] = alloca [9 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS13:%.*]] = alloca [9 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS14:%.*]] = alloca [9 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS14:%.*]] = alloca [9 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS15:%.*]] = alloca [9 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS15:%.*]] = alloca [9 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]]) // CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
// CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK4-NEXT: store i32 0, i32* [[A]], align 4 // CHECK4-NEXT: store i32 0, i32* [[A]], align 4
// CHECK4-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK4-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK4-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK4-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 85 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP47:%.*]] = mul nuw i32 [[TMP46]], 8 // CHECK4-NEXT: [[TMP47:%.*]] = mul nuw i32 [[TMP46]], 8
// CHECK4-NEXT: [[TMP48:%.*]] = sext i32 [[TMP47]] to i64 // CHECK4-NEXT: [[TMP48:%.*]] = sext i32 [[TMP47]] to i64
// CHECK4-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0 // CHECK4-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK4-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32* // CHECK4-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32*
// CHECK4-NEXT: store i32 [[TMP42]], i32* [[TMP50]], align 4 // CHECK4-NEXT: store i32 [[TMP42]], i32* [[TMP50]], align 4
// CHECK4-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0 // CHECK4-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK4-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32* // CHECK4-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32*
// CHECK4-NEXT: store i32 [[TMP42]], i32* [[TMP52]], align 4 // CHECK4-NEXT: store i32 [[TMP42]], i32* [[TMP52]], align 4
// CHECK4-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK4-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 0
// CHECK4-NEXT: store i64 4, i64* [[TMP53]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP53]], align 4
// CHECK4-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 0 // CHECK4-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 1
// CHECK4-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK4-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to [10 x float]**
// CHECK4-NEXT: [[TMP55:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 1 // CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP55]], align 4
// CHECK4-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to [10 x float]** // CHECK4-NEXT: [[TMP56:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 1
// CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP56]], align 4 // CHECK4-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to [10 x float]**
// CHECK4-NEXT: [[TMP57:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 1 // CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP57]], align 4
// CHECK4-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to [10 x float]** // CHECK4-NEXT: [[TMP58:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 1
// CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP58]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP58]], align 4
// CHECK4-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK4-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 2
// CHECK4-NEXT: store i64 40, i64* [[TMP59]], align 4 // CHECK4-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i32*
// CHECK4-NEXT: [[TMP60:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP60]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP60]], align 4 // CHECK4-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 2
// CHECK4-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 2
// CHECK4-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i32* // CHECK4-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i32*
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP62]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP62]], align 4
// CHECK4-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 2 // CHECK4-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 2
// CHECK4-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32* // CHECK4-NEXT: store i8* null, i8** [[TMP63]], align 4
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP64]], align 4 // CHECK4-NEXT: [[TMP64:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 3
// CHECK4-NEXT: [[TMP65:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK4-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to float**
// CHECK4-NEXT: store i64 4, i64* [[TMP65]], align 4 // CHECK4-NEXT: store float* [[VLA]], float** [[TMP65]], align 4
// CHECK4-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 2 // CHECK4-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP66]], align 4 // CHECK4-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to float**
// CHECK4-NEXT: [[TMP67:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 3 // CHECK4-NEXT: store float* [[VLA]], float** [[TMP67]], align 4
// CHECK4-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to float** // CHECK4-NEXT: store i64 [[TMP45]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK4-NEXT: store float* [[VLA]], float** [[TMP68]], align 4 // CHECK4-NEXT: [[TMP68:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 3
// CHECK4-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 3 // CHECK4-NEXT: store i8* null, i8** [[TMP68]], align 4
// CHECK4-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to float** // CHECK4-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 4
// CHECK4-NEXT: store float* [[VLA]], float** [[TMP70]], align 4 // CHECK4-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to [5 x [10 x double]]**
// CHECK4-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP70]], align 4
// CHECK4-NEXT: store i64 [[TMP45]], i64* [[TMP71]], align 4 // CHECK4-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 4
// CHECK4-NEXT: [[TMP72:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 3 // CHECK4-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [5 x [10 x double]]**
// CHECK4-NEXT: store i8* null, i8** [[TMP72]], align 4 // CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP72]], align 4
// CHECK4-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 4 // CHECK4-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 4
// CHECK4-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [5 x [10 x double]]** // CHECK4-NEXT: store i8* null, i8** [[TMP73]], align 4
// CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP74]], align 4 // CHECK4-NEXT: [[TMP74:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 5
// CHECK4-NEXT: [[TMP75:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 4 // CHECK4-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32*
// CHECK4-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [5 x [10 x double]]** // CHECK4-NEXT: store i32 5, i32* [[TMP75]], align 4
// CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP76]], align 4 // CHECK4-NEXT: [[TMP76:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 5
// CHECK4-NEXT: [[TMP77:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK4-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32*
// CHECK4-NEXT: store i64 400, i64* [[TMP77]], align 4 // CHECK4-NEXT: store i32 5, i32* [[TMP77]], align 4
// CHECK4-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 4 // CHECK4-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 5
// CHECK4-NEXT: store i8* null, i8** [[TMP78]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP78]], align 4
// CHECK4-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 5 // CHECK4-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 6
// CHECK4-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK4-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32*
// CHECK4-NEXT: store i32 5, i32* [[TMP80]], align 4 // CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP80]], align 4
// CHECK4-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 5 // CHECK4-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 6
// CHECK4-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32* // CHECK4-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32*
// CHECK4-NEXT: store i32 5, i32* [[TMP82]], align 4 // CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP82]], align 4
// CHECK4-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK4-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 6
// CHECK4-NEXT: store i64 4, i64* [[TMP83]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP83]], align 4
// CHECK4-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 5 // CHECK4-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 7
// CHECK4-NEXT: store i8* null, i8** [[TMP84]], align 4 // CHECK4-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to double**
// CHECK4-NEXT: [[TMP85:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 6 // CHECK4-NEXT: store double* [[VLA1]], double** [[TMP85]], align 4
// CHECK4-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32* // CHECK4-NEXT: [[TMP86:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 7
// CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP86]], align 4 // CHECK4-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to double**
// CHECK4-NEXT: [[TMP87:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 6 // CHECK4-NEXT: store double* [[VLA1]], double** [[TMP87]], align 4
// CHECK4-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32* // CHECK4-NEXT: store i64 [[TMP48]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP88]], align 4 // CHECK4-NEXT: [[TMP88:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 7
// CHECK4-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK4-NEXT: store i8* null, i8** [[TMP88]], align 4
// CHECK4-NEXT: store i64 4, i64* [[TMP89]], align 4 // CHECK4-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 8
// CHECK4-NEXT: [[TMP90:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 6 // CHECK4-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to %struct.TT**
// CHECK4-NEXT: store i8* null, i8** [[TMP90]], align 4 // CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP90]], align 4
// CHECK4-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 7 // CHECK4-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 8
// CHECK4-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to double** // CHECK4-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to %struct.TT**
// CHECK4-NEXT: store double* [[VLA1]], double** [[TMP92]], align 4 // CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP92]], align 4
// CHECK4-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 7 // CHECK4-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 8
// CHECK4-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to double** // CHECK4-NEXT: store i8* null, i8** [[TMP93]], align 4
// CHECK4-NEXT: store double* [[VLA1]], double** [[TMP94]], align 4 // CHECK4-NEXT: [[TMP94:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK4-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK4-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK4-NEXT: store i64 [[TMP48]], i64* [[TMP95]], align 4 // CHECK4-NEXT: [[TMP96:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP94]], i8** [[TMP95]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP96:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 7 // CHECK4-NEXT: [[TMP97:%.*]] = icmp ne i32 [[TMP96]], 0
// CHECK4-NEXT: store i8* null, i8** [[TMP96]], align 4 // CHECK4-NEXT: br i1 [[TMP97]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK4-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 8
// CHECK4-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to %struct.TT**
// CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP98]], align 4
// CHECK4-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 8
// CHECK4-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to %struct.TT**
// CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP100]], align 4
// CHECK4-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK4-NEXT: store i64 12, i64* [[TMP101]], align 4
// CHECK4-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 8
// CHECK4-NEXT: store i8* null, i8** [[TMP102]], align 4
// CHECK4-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK4-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK4-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK4-NEXT: [[TMP106:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP103]], i8** [[TMP104]], i64* [[TMP105]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP107:%.*]] = icmp ne i32 [[TMP106]], 0
// CHECK4-NEXT: br i1 [[TMP107]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK4: omp_offload.failed16: // CHECK4: omp_offload.failed16:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT17]]
// CHECK4: omp_offload.cont17: // CHECK4: omp_offload.cont17:
// CHECK4-NEXT: br label [[OMP_IF_END19:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END19:%.*]]
// CHECK4: omp_if.else18: // CHECK4: omp_if.else18:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_IF_END19]] // CHECK4-NEXT: br label [[OMP_IF_END19]]
// CHECK4: omp_if.end19: // CHECK4: omp_if.end19:
// CHECK4-NEXT: [[TMP108:%.*]] = load i32, i32* [[A]], align 4 // CHECK4-NEXT: [[TMP98:%.*]] = load i32, i32* [[A]], align 4
// CHECK4-NEXT: [[TMP109:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: [[TMP99:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP109]]) // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP99]])
// CHECK4-NEXT: ret i32 [[TMP108]] // CHECK4-NEXT: ret i32 [[TMP98]]
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100
// CHECK4-SAME: () #[[ATTR2:[0-9]+]] { // CHECK4-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
▲ Show 20 Lines • Show All 312 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK4-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK4-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK4-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 13 Lines
// CHECK4-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK4-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK4-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK4-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK4-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK4-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK4-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK4-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK4-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK4-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK4-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK4-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK4-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK4-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK4-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK4-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK4-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK4-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK4-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK4-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK4-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK4-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK4-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK4-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK4-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK4-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK4-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK4-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK4-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK4-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK4-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK4-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK4-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK4-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK4-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK4-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK4-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK4-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK4-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK4-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK4-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK4-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK4-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK4-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK4-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4 // CHECK4-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK4-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK4-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK4-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK4-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK4-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK4-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK4-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_IF_END]] // CHECK4-NEXT: br label [[OMP_IF_END]]
// CHECK4: omp_if.end: // CHECK4: omp_if.end:
// CHECK4-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK4-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK4-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK4-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK4-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK4-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK4-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK4-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK4-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK4-NEXT: ret i32 [[ADD3]] // CHECK4-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK4-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK4-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK4-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK4-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK4-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK4-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK4-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK4-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK4-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
Show All 20 Lines
// CHECK4-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK4-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK4-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK4-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 49 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK4-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK4-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK4-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK4-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK4-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK4-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK4-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK4-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
▲ Show 20 Lines • Show All 41 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK4-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK4-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK4-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
▲ Show 20 Lines • Show All 1,673 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[AA_CASTED5:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AA_CASTED5:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS7:%.*]] = alloca [2 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS7:%.*]] = alloca [2 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS8:%.*]] = alloca [2 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS8:%.*]] = alloca [2 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS9:%.*]] = alloca [2 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS9:%.*]] = alloca [2 x i8*], align 8
// CHECK17-NEXT: [[A_CASTED12:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[A_CASTED12:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [9 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [9 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [9 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [9 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [9 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [9 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]]) // CHECK17-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
// CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK17-NEXT: store i32 0, i32* [[A]], align 4 // CHECK17-NEXT: store i32 0, i32* [[A]], align 4
// CHECK17-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK17-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK17-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK17-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK17-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK17-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK17-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[TMP47:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK17-NEXT: [[TMP47:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK17-NEXT: [[TMP48:%.*]] = mul nuw i64 [[TMP47]], 8 // CHECK17-NEXT: [[TMP48:%.*]] = mul nuw i64 [[TMP47]], 8
// CHECK17-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0 // CHECK17-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK17-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i64* // CHECK17-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i64*
// CHECK17-NEXT: store i64 [[TMP44]], i64* [[TMP50]], align 8 // CHECK17-NEXT: store i64 [[TMP44]], i64* [[TMP50]], align 8
// CHECK17-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0 // CHECK17-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK17-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i64* // CHECK17-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i64*
// CHECK17-NEXT: store i64 [[TMP44]], i64* [[TMP52]], align 8 // CHECK17-NEXT: store i64 [[TMP44]], i64* [[TMP52]], align 8
// CHECK17-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK17-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 0
// CHECK17-NEXT: store i64 4, i64* [[TMP53]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP53]], align 8
// CHECK17-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 0 // CHECK17-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 1
// CHECK17-NEXT: store i8* null, i8** [[TMP54]], align 8 // CHECK17-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to [10 x float]**
// CHECK17-NEXT: [[TMP55:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 1 // CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP55]], align 8
// CHECK17-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to [10 x float]** // CHECK17-NEXT: [[TMP56:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 1
// CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP56]], align 8 // CHECK17-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to [10 x float]**
// CHECK17-NEXT: [[TMP57:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 1 // CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP57]], align 8
// CHECK17-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to [10 x float]** // CHECK17-NEXT: [[TMP58:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 1
// CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP58]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP58]], align 8
// CHECK17-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK17-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 2
// CHECK17-NEXT: store i64 40, i64* [[TMP59]], align 8 // CHECK17-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i64*
// CHECK17-NEXT: [[TMP60:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 1 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP60]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP60]], align 8 // CHECK17-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 2
// CHECK17-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 2
// CHECK17-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64* // CHECK17-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP62]], align 8 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP62]], align 8
// CHECK17-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 2 // CHECK17-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 2
// CHECK17-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64* // CHECK17-NEXT: store i8* null, i8** [[TMP63]], align 8
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP64]], align 8 // CHECK17-NEXT: [[TMP64:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 3
// CHECK17-NEXT: [[TMP65:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK17-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to float**
// CHECK17-NEXT: store i64 8, i64* [[TMP65]], align 8 // CHECK17-NEXT: store float* [[VLA]], float** [[TMP65]], align 8
// CHECK17-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 2 // CHECK17-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 3
// CHECK17-NEXT: store i8* null, i8** [[TMP66]], align 8 // CHECK17-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to float**
// CHECK17-NEXT: [[TMP67:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 3 // CHECK17-NEXT: store float* [[VLA]], float** [[TMP67]], align 8
// CHECK17-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to float** // CHECK17-NEXT: store i64 [[TMP46]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK17-NEXT: store float* [[VLA]], float** [[TMP68]], align 8 // CHECK17-NEXT: [[TMP68:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 3
// CHECK17-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 3 // CHECK17-NEXT: store i8* null, i8** [[TMP68]], align 8
// CHECK17-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to float** // CHECK17-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 4
// CHECK17-NEXT: store float* [[VLA]], float** [[TMP70]], align 8 // CHECK17-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to [5 x [10 x double]]**
// CHECK17-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP70]], align 8
// CHECK17-NEXT: store i64 [[TMP46]], i64* [[TMP71]], align 8 // CHECK17-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 4
// CHECK17-NEXT: [[TMP72:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 3 // CHECK17-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [5 x [10 x double]]**
// CHECK17-NEXT: store i8* null, i8** [[TMP72]], align 8 // CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP72]], align 8
// CHECK17-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 4 // CHECK17-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 4
// CHECK17-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [5 x [10 x double]]** // CHECK17-NEXT: store i8* null, i8** [[TMP73]], align 8
// CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP74]], align 8 // CHECK17-NEXT: [[TMP74:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 5
// CHECK17-NEXT: [[TMP75:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 4 // CHECK17-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64*
// CHECK17-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [5 x [10 x double]]** // CHECK17-NEXT: store i64 5, i64* [[TMP75]], align 8
// CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP76]], align 8 // CHECK17-NEXT: [[TMP76:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 5
// CHECK17-NEXT: [[TMP77:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK17-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK17-NEXT: store i64 400, i64* [[TMP77]], align 8 // CHECK17-NEXT: store i64 5, i64* [[TMP77]], align 8
// CHECK17-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 4 // CHECK17-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 5
// CHECK17-NEXT: store i8* null, i8** [[TMP78]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP78]], align 8
// CHECK17-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 5 // CHECK17-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 6
// CHECK17-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i64* // CHECK17-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i64*
// CHECK17-NEXT: store i64 5, i64* [[TMP80]], align 8 // CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP80]], align 8
// CHECK17-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 5 // CHECK17-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 6
// CHECK17-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i64* // CHECK17-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i64*
// CHECK17-NEXT: store i64 5, i64* [[TMP82]], align 8 // CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP82]], align 8
// CHECK17-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK17-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 6
// CHECK17-NEXT: store i64 8, i64* [[TMP83]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP83]], align 8
// CHECK17-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 5 // CHECK17-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 7
// CHECK17-NEXT: store i8* null, i8** [[TMP84]], align 8 // CHECK17-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to double**
// CHECK17-NEXT: [[TMP85:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 6 // CHECK17-NEXT: store double* [[VLA1]], double** [[TMP85]], align 8
// CHECK17-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i64* // CHECK17-NEXT: [[TMP86:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 7
// CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP86]], align 8 // CHECK17-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to double**
// CHECK17-NEXT: [[TMP87:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 6 // CHECK17-NEXT: store double* [[VLA1]], double** [[TMP87]], align 8
// CHECK17-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK17-NEXT: store i64 [[TMP48]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP88]], align 8 // CHECK17-NEXT: [[TMP88:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 7
// CHECK17-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK17-NEXT: store i8* null, i8** [[TMP88]], align 8
// CHECK17-NEXT: store i64 8, i64* [[TMP89]], align 8 // CHECK17-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 8
// CHECK17-NEXT: [[TMP90:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 6 // CHECK17-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to %struct.TT**
// CHECK17-NEXT: store i8* null, i8** [[TMP90]], align 8 // CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP90]], align 8
// CHECK17-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 7 // CHECK17-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 8
// CHECK17-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to double** // CHECK17-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to %struct.TT**
// CHECK17-NEXT: store double* [[VLA1]], double** [[TMP92]], align 8 // CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP92]], align 8
// CHECK17-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 7 // CHECK17-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 8
// CHECK17-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to double** // CHECK17-NEXT: store i8* null, i8** [[TMP93]], align 8
// CHECK17-NEXT: store double* [[VLA1]], double** [[TMP94]], align 8 // CHECK17-NEXT: [[TMP94:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK17-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK17-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK17-NEXT: store i64 [[TMP48]], i64* [[TMP95]], align 8 // CHECK17-NEXT: [[TMP96:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP94]], i8** [[TMP95]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: [[TMP96:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 7 // CHECK17-NEXT: [[TMP97:%.*]] = icmp ne i32 [[TMP96]], 0
// CHECK17-NEXT: store i8* null, i8** [[TMP96]], align 8 // CHECK17-NEXT: br i1 [[TMP97]], label [[OMP_OFFLOAD_FAILED19:%.*]], label [[OMP_OFFLOAD_CONT20:%.*]]
// CHECK17-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 8
// CHECK17-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to %struct.TT**
// CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP98]], align 8
// CHECK17-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 8
// CHECK17-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to %struct.TT**
// CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP100]], align 8
// CHECK17-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK17-NEXT: store i64 16, i64* [[TMP101]], align 8
// CHECK17-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 8
// CHECK17-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK17-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK17-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK17-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK17-NEXT: [[TMP106:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP103]], i8** [[TMP104]], i64* [[TMP105]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: [[TMP107:%.*]] = icmp ne i32 [[TMP106]], 0
// CHECK17-NEXT: br i1 [[TMP107]], label [[OMP_OFFLOAD_FAILED19:%.*]], label [[OMP_OFFLOAD_CONT20:%.*]]
// CHECK17: omp_offload.failed19: // CHECK17: omp_offload.failed19:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT20]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT20]]
// CHECK17: omp_offload.cont20: // CHECK17: omp_offload.cont20:
// CHECK17-NEXT: br label [[OMP_IF_END22:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END22:%.*]]
// CHECK17: omp_if.else21: // CHECK17: omp_if.else21:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_IF_END22]] // CHECK17-NEXT: br label [[OMP_IF_END22]]
// CHECK17: omp_if.end22: // CHECK17: omp_if.end22:
// CHECK17-NEXT: [[TMP108:%.*]] = load i32, i32* [[A]], align 4 // CHECK17-NEXT: [[TMP98:%.*]] = load i32, i32* [[A]], align 4
// CHECK17-NEXT: [[TMP109:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK17-NEXT: [[TMP99:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP109]]) // CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP99]])
// CHECK17-NEXT: ret i32 [[TMP108]] // CHECK17-NEXT: ret i32 [[TMP98]]
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100
// CHECK17-SAME: () #[[ATTR2:[0-9]+]] { // CHECK17-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
▲ Show 20 Lines • Show All 321 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK17-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK17-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK17-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK17-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK17-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK17-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 14 Lines
// CHECK17-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK17-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK17-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK17-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK17-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK17-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK17-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK17-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK17-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK17-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK17-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK17-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK17-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK17-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK17-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK17-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK17-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK17-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK17-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK17-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK17-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK17-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK17-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK17-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK17-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK17-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK17-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK17-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK17-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK17-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK17-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK17-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK17-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK17-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK17-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK17-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK17-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK17-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK17-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK17-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK17-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK17-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8 // CHECK17-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK17-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK17-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK17-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK17-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK17-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK17-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK17-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: br label [[OMP_IF_END:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK17: omp_if.else: // CHECK17: omp_if.else:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_IF_END]] // CHECK17-NEXT: br label [[OMP_IF_END]]
// CHECK17: omp_if.end: // CHECK17: omp_if.end:
// CHECK17-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK17-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK17-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK17-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK17-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK17-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK17-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK17-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK17-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK17-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK17-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK17-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK17-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK17-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK17-NEXT: ret i32 [[ADD4]] // CHECK17-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK17-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK17-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK17-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK17-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK17-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK17-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK17-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK17-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK17-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK17-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: br label [[OMP_IF_END:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK17: omp_if.else: // CHECK17: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK17-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK17-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK17-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK17-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK17-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: br label [[OMP_IF_END:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK17: omp_if.else: // CHECK17: omp_if.else:
Show All 22 Lines
// CHECK17-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK17-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK17-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK17-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK17-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK17-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK17-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK17-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK17-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK17-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK17-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK17-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK17-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK17-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK17-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK17-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK17-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK17-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK17-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK17-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK17-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK17-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK17-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK17-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK17-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK17-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK17-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK17-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK17-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK17-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK17-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK17-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK17-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK17-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK17-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK17-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[AA_CASTED5:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AA_CASTED5:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS7:%.*]] = alloca [2 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS7:%.*]] = alloca [2 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS8:%.*]] = alloca [2 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS8:%.*]] = alloca [2 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS9:%.*]] = alloca [2 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS9:%.*]] = alloca [2 x i8*], align 8
// CHECK18-NEXT: [[A_CASTED12:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[A_CASTED12:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [9 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [9 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [9 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [9 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [9 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [9 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 8
// CHECK18-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]]) // CHECK18-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
// CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK18-NEXT: store i32 0, i32* [[A]], align 4 // CHECK18-NEXT: store i32 0, i32* [[A]], align 4
// CHECK18-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK18-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK18-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK18-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK18-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK18-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK18-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[TMP47:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK18-NEXT: [[TMP47:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK18-NEXT: [[TMP48:%.*]] = mul nuw i64 [[TMP47]], 8 // CHECK18-NEXT: [[TMP48:%.*]] = mul nuw i64 [[TMP47]], 8
// CHECK18-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0 // CHECK18-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK18-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i64* // CHECK18-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i64*
// CHECK18-NEXT: store i64 [[TMP44]], i64* [[TMP50]], align 8 // CHECK18-NEXT: store i64 [[TMP44]], i64* [[TMP50]], align 8
// CHECK18-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0 // CHECK18-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK18-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i64* // CHECK18-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i64*
// CHECK18-NEXT: store i64 [[TMP44]], i64* [[TMP52]], align 8 // CHECK18-NEXT: store i64 [[TMP44]], i64* [[TMP52]], align 8
// CHECK18-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK18-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 0
// CHECK18-NEXT: store i64 4, i64* [[TMP53]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP53]], align 8
// CHECK18-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 0 // CHECK18-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 1
// CHECK18-NEXT: store i8* null, i8** [[TMP54]], align 8 // CHECK18-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to [10 x float]**
// CHECK18-NEXT: [[TMP55:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 1 // CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP55]], align 8
// CHECK18-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to [10 x float]** // CHECK18-NEXT: [[TMP56:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 1
// CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP56]], align 8 // CHECK18-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to [10 x float]**
// CHECK18-NEXT: [[TMP57:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 1 // CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP57]], align 8
// CHECK18-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to [10 x float]** // CHECK18-NEXT: [[TMP58:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 1
// CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP58]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP58]], align 8
// CHECK18-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK18-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 2
// CHECK18-NEXT: store i64 40, i64* [[TMP59]], align 8 // CHECK18-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i64*
// CHECK18-NEXT: [[TMP60:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 1 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP60]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP60]], align 8 // CHECK18-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 2
// CHECK18-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 2
// CHECK18-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64* // CHECK18-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP62]], align 8 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP62]], align 8
// CHECK18-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 2 // CHECK18-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 2
// CHECK18-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64* // CHECK18-NEXT: store i8* null, i8** [[TMP63]], align 8
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP64]], align 8 // CHECK18-NEXT: [[TMP64:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 3
// CHECK18-NEXT: [[TMP65:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK18-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to float**
// CHECK18-NEXT: store i64 8, i64* [[TMP65]], align 8 // CHECK18-NEXT: store float* [[VLA]], float** [[TMP65]], align 8
// CHECK18-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 2 // CHECK18-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 3
// CHECK18-NEXT: store i8* null, i8** [[TMP66]], align 8 // CHECK18-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to float**
// CHECK18-NEXT: [[TMP67:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 3 // CHECK18-NEXT: store float* [[VLA]], float** [[TMP67]], align 8
// CHECK18-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to float** // CHECK18-NEXT: store i64 [[TMP46]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK18-NEXT: store float* [[VLA]], float** [[TMP68]], align 8 // CHECK18-NEXT: [[TMP68:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 3
// CHECK18-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 3 // CHECK18-NEXT: store i8* null, i8** [[TMP68]], align 8
// CHECK18-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to float** // CHECK18-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 4
// CHECK18-NEXT: store float* [[VLA]], float** [[TMP70]], align 8 // CHECK18-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to [5 x [10 x double]]**
// CHECK18-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP70]], align 8
// CHECK18-NEXT: store i64 [[TMP46]], i64* [[TMP71]], align 8 // CHECK18-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 4
// CHECK18-NEXT: [[TMP72:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 3 // CHECK18-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [5 x [10 x double]]**
// CHECK18-NEXT: store i8* null, i8** [[TMP72]], align 8 // CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP72]], align 8
// CHECK18-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 4 // CHECK18-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 4
// CHECK18-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [5 x [10 x double]]** // CHECK18-NEXT: store i8* null, i8** [[TMP73]], align 8
// CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP74]], align 8 // CHECK18-NEXT: [[TMP74:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 5
// CHECK18-NEXT: [[TMP75:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 4 // CHECK18-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64*
// CHECK18-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [5 x [10 x double]]** // CHECK18-NEXT: store i64 5, i64* [[TMP75]], align 8
// CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP76]], align 8 // CHECK18-NEXT: [[TMP76:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 5
// CHECK18-NEXT: [[TMP77:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK18-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK18-NEXT: store i64 400, i64* [[TMP77]], align 8 // CHECK18-NEXT: store i64 5, i64* [[TMP77]], align 8
// CHECK18-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 4 // CHECK18-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 5
// CHECK18-NEXT: store i8* null, i8** [[TMP78]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP78]], align 8
// CHECK18-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 5 // CHECK18-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 6
// CHECK18-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i64* // CHECK18-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i64*
// CHECK18-NEXT: store i64 5, i64* [[TMP80]], align 8 // CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP80]], align 8
// CHECK18-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 5 // CHECK18-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 6
// CHECK18-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i64* // CHECK18-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i64*
// CHECK18-NEXT: store i64 5, i64* [[TMP82]], align 8 // CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP82]], align 8
// CHECK18-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK18-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 6
// CHECK18-NEXT: store i64 8, i64* [[TMP83]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP83]], align 8
// CHECK18-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 5 // CHECK18-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 7
// CHECK18-NEXT: store i8* null, i8** [[TMP84]], align 8 // CHECK18-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to double**
// CHECK18-NEXT: [[TMP85:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 6 // CHECK18-NEXT: store double* [[VLA1]], double** [[TMP85]], align 8
// CHECK18-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i64* // CHECK18-NEXT: [[TMP86:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 7
// CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP86]], align 8 // CHECK18-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to double**
// CHECK18-NEXT: [[TMP87:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 6 // CHECK18-NEXT: store double* [[VLA1]], double** [[TMP87]], align 8
// CHECK18-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK18-NEXT: store i64 [[TMP48]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP88]], align 8 // CHECK18-NEXT: [[TMP88:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 7
// CHECK18-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK18-NEXT: store i8* null, i8** [[TMP88]], align 8
// CHECK18-NEXT: store i64 8, i64* [[TMP89]], align 8 // CHECK18-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 8
// CHECK18-NEXT: [[TMP90:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 6 // CHECK18-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to %struct.TT**
// CHECK18-NEXT: store i8* null, i8** [[TMP90]], align 8 // CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP90]], align 8
// CHECK18-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 7 // CHECK18-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 8
// CHECK18-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to double** // CHECK18-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to %struct.TT**
// CHECK18-NEXT: store double* [[VLA1]], double** [[TMP92]], align 8 // CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP92]], align 8
// CHECK18-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 7 // CHECK18-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 8
// CHECK18-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to double** // CHECK18-NEXT: store i8* null, i8** [[TMP93]], align 8
// CHECK18-NEXT: store double* [[VLA1]], double** [[TMP94]], align 8 // CHECK18-NEXT: [[TMP94:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK18-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK18-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK18-NEXT: store i64 [[TMP48]], i64* [[TMP95]], align 8 // CHECK18-NEXT: [[TMP96:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP94]], i8** [[TMP95]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: [[TMP96:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 7 // CHECK18-NEXT: [[TMP97:%.*]] = icmp ne i32 [[TMP96]], 0
// CHECK18-NEXT: store i8* null, i8** [[TMP96]], align 8 // CHECK18-NEXT: br i1 [[TMP97]], label [[OMP_OFFLOAD_FAILED19:%.*]], label [[OMP_OFFLOAD_CONT20:%.*]]
// CHECK18-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 8
// CHECK18-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to %struct.TT**
// CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP98]], align 8
// CHECK18-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 8
// CHECK18-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to %struct.TT**
// CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP100]], align 8
// CHECK18-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK18-NEXT: store i64 16, i64* [[TMP101]], align 8
// CHECK18-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS18]], i64 0, i64 8
// CHECK18-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK18-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS16]], i32 0, i32 0
// CHECK18-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS17]], i32 0, i32 0
// CHECK18-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK18-NEXT: [[TMP106:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP103]], i8** [[TMP104]], i64* [[TMP105]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: [[TMP107:%.*]] = icmp ne i32 [[TMP106]], 0
// CHECK18-NEXT: br i1 [[TMP107]], label [[OMP_OFFLOAD_FAILED19:%.*]], label [[OMP_OFFLOAD_CONT20:%.*]]
// CHECK18: omp_offload.failed19: // CHECK18: omp_offload.failed19:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT20]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT20]]
// CHECK18: omp_offload.cont20: // CHECK18: omp_offload.cont20:
// CHECK18-NEXT: br label [[OMP_IF_END22:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END22:%.*]]
// CHECK18: omp_if.else21: // CHECK18: omp_if.else21:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i64 [[TMP44]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_IF_END22]] // CHECK18-NEXT: br label [[OMP_IF_END22]]
// CHECK18: omp_if.end22: // CHECK18: omp_if.end22:
// CHECK18-NEXT: [[TMP108:%.*]] = load i32, i32* [[A]], align 4 // CHECK18-NEXT: [[TMP98:%.*]] = load i32, i32* [[A]], align 4
// CHECK18-NEXT: [[TMP109:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK18-NEXT: [[TMP99:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP109]]) // CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP99]])
// CHECK18-NEXT: ret i32 [[TMP108]] // CHECK18-NEXT: ret i32 [[TMP98]]
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100 // CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100
// CHECK18-SAME: () #[[ATTR2:[0-9]+]] { // CHECK18-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
▲ Show 20 Lines • Show All 321 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK18-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK18-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK18-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK18-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK18-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK18-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK18-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK18-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK18-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK18-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 14 Lines
// CHECK18-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK18-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK18-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK18-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK18-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK18-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK18-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK18-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK18-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK18-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK18-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK18-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK18-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK18-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK18-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK18-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK18-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK18-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK18-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK18-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK18-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK18-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK18-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK18-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK18-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK18-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK18-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK18-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK18-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK18-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK18-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK18-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK18-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK18-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK18-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK18-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK18-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK18-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK18-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK18-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK18-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK18-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK18-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK18-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK18-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK18-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK18-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK18-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK18-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK18-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK18-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK18-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK18-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK18-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK18-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK18-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK18-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8 // CHECK18-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK18-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK18-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK18-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK18-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK18-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK18-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK18-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: br label [[OMP_IF_END:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK18: omp_if.else: // CHECK18: omp_if.else:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_IF_END]] // CHECK18-NEXT: br label [[OMP_IF_END]]
// CHECK18: omp_if.end: // CHECK18: omp_if.end:
// CHECK18-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK18-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK18-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK18-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK18-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK18-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK18-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK18-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK18-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK18-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK18-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK18-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK18-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK18-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK18-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK18-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK18-NEXT: ret i32 [[ADD4]] // CHECK18-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK18-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK18-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK18-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK18-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK18-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK18-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK18-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK18-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK18-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK18-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: br label [[OMP_IF_END:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK18: omp_if.else: // CHECK18: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK18-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK18-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK18-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK18-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK18-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: br label [[OMP_IF_END:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK18: omp_if.else: // CHECK18: omp_if.else:
Show All 22 Lines
// CHECK18-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK18-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK18-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK18-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK18-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK18-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK18-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK18-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK18-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK18-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK18-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK18-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK18-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK18-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK18-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK18-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK18-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK18-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK18-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK18-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK18-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK18-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK18-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK18-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK18-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK18-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK18-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK18-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK18-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK18-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK18-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK18-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK18-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK18-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK18-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK18-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK18-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK18-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK18-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK18-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK18-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK18-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK18-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[AA_CASTED3:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AA_CASTED3:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4
// CHECK19-NEXT: [[A_CASTED10:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A_CASTED10:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS13:%.*]] = alloca [9 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS13:%.*]] = alloca [9 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS14:%.*]] = alloca [9 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS14:%.*]] = alloca [9 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS15:%.*]] = alloca [9 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS15:%.*]] = alloca [9 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]]) // CHECK19-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
// CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK19-NEXT: store i32 0, i32* [[A]], align 4 // CHECK19-NEXT: store i32 0, i32* [[A]], align 4
// CHECK19-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK19-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK19-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK19-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK19-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK19-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 85 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[TMP47:%.*]] = mul nuw i32 [[TMP46]], 8 // CHECK19-NEXT: [[TMP47:%.*]] = mul nuw i32 [[TMP46]], 8
// CHECK19-NEXT: [[TMP48:%.*]] = sext i32 [[TMP47]] to i64 // CHECK19-NEXT: [[TMP48:%.*]] = sext i32 [[TMP47]] to i64
// CHECK19-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0 // CHECK19-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK19-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32* // CHECK19-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32*
// CHECK19-NEXT: store i32 [[TMP42]], i32* [[TMP50]], align 4 // CHECK19-NEXT: store i32 [[TMP42]], i32* [[TMP50]], align 4
// CHECK19-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0 // CHECK19-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK19-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32* // CHECK19-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32*
// CHECK19-NEXT: store i32 [[TMP42]], i32* [[TMP52]], align 4 // CHECK19-NEXT: store i32 [[TMP42]], i32* [[TMP52]], align 4
// CHECK19-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK19-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 0
// CHECK19-NEXT: store i64 4, i64* [[TMP53]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP53]], align 4
// CHECK19-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 0 // CHECK19-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 1
// CHECK19-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK19-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to [10 x float]**
// CHECK19-NEXT: [[TMP55:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 1 // CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP55]], align 4
// CHECK19-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to [10 x float]** // CHECK19-NEXT: [[TMP56:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 1
// CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP56]], align 4 // CHECK19-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to [10 x float]**
// CHECK19-NEXT: [[TMP57:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 1 // CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP57]], align 4
// CHECK19-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to [10 x float]** // CHECK19-NEXT: [[TMP58:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 1
// CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP58]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP58]], align 4
// CHECK19-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK19-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 2
// CHECK19-NEXT: store i64 40, i64* [[TMP59]], align 4 // CHECK19-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i32*
// CHECK19-NEXT: [[TMP60:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP60]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP60]], align 4 // CHECK19-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 2
// CHECK19-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 2
// CHECK19-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i32* // CHECK19-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i32*
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP62]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP62]], align 4
// CHECK19-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 2 // CHECK19-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 2
// CHECK19-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32* // CHECK19-NEXT: store i8* null, i8** [[TMP63]], align 4
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP64]], align 4 // CHECK19-NEXT: [[TMP64:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 3
// CHECK19-NEXT: [[TMP65:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK19-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to float**
// CHECK19-NEXT: store i64 4, i64* [[TMP65]], align 4 // CHECK19-NEXT: store float* [[VLA]], float** [[TMP65]], align 4
// CHECK19-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 2 // CHECK19-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 3
// CHECK19-NEXT: store i8* null, i8** [[TMP66]], align 4 // CHECK19-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to float**
// CHECK19-NEXT: [[TMP67:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 3 // CHECK19-NEXT: store float* [[VLA]], float** [[TMP67]], align 4
// CHECK19-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to float** // CHECK19-NEXT: store i64 [[TMP45]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK19-NEXT: store float* [[VLA]], float** [[TMP68]], align 4 // CHECK19-NEXT: [[TMP68:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 3
// CHECK19-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 3 // CHECK19-NEXT: store i8* null, i8** [[TMP68]], align 4
// CHECK19-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to float** // CHECK19-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 4
// CHECK19-NEXT: store float* [[VLA]], float** [[TMP70]], align 4 // CHECK19-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to [5 x [10 x double]]**
// CHECK19-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP70]], align 4
// CHECK19-NEXT: store i64 [[TMP45]], i64* [[TMP71]], align 4 // CHECK19-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 4
// CHECK19-NEXT: [[TMP72:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 3 // CHECK19-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [5 x [10 x double]]**
// CHECK19-NEXT: store i8* null, i8** [[TMP72]], align 4 // CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP72]], align 4
// CHECK19-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 4 // CHECK19-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 4
// CHECK19-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [5 x [10 x double]]** // CHECK19-NEXT: store i8* null, i8** [[TMP73]], align 4
// CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP74]], align 4 // CHECK19-NEXT: [[TMP74:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 5
// CHECK19-NEXT: [[TMP75:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 4 // CHECK19-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32*
// CHECK19-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [5 x [10 x double]]** // CHECK19-NEXT: store i32 5, i32* [[TMP75]], align 4
// CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP76]], align 4 // CHECK19-NEXT: [[TMP76:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 5
// CHECK19-NEXT: [[TMP77:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK19-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32*
// CHECK19-NEXT: store i64 400, i64* [[TMP77]], align 4 // CHECK19-NEXT: store i32 5, i32* [[TMP77]], align 4
// CHECK19-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 4 // CHECK19-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 5
// CHECK19-NEXT: store i8* null, i8** [[TMP78]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP78]], align 4
// CHECK19-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 5 // CHECK19-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 6
// CHECK19-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK19-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32*
// CHECK19-NEXT: store i32 5, i32* [[TMP80]], align 4 // CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP80]], align 4
// CHECK19-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 5 // CHECK19-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 6
// CHECK19-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32* // CHECK19-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32*
// CHECK19-NEXT: store i32 5, i32* [[TMP82]], align 4 // CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP82]], align 4
// CHECK19-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK19-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 6
// CHECK19-NEXT: store i64 4, i64* [[TMP83]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP83]], align 4
// CHECK19-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 5 // CHECK19-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 7
// CHECK19-NEXT: store i8* null, i8** [[TMP84]], align 4 // CHECK19-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to double**
// CHECK19-NEXT: [[TMP85:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 6 // CHECK19-NEXT: store double* [[VLA1]], double** [[TMP85]], align 4
// CHECK19-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32* // CHECK19-NEXT: [[TMP86:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 7
// CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP86]], align 4 // CHECK19-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to double**
// CHECK19-NEXT: [[TMP87:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 6 // CHECK19-NEXT: store double* [[VLA1]], double** [[TMP87]], align 4
// CHECK19-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32* // CHECK19-NEXT: store i64 [[TMP48]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP88]], align 4 // CHECK19-NEXT: [[TMP88:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 7
// CHECK19-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK19-NEXT: store i8* null, i8** [[TMP88]], align 4
// CHECK19-NEXT: store i64 4, i64* [[TMP89]], align 4 // CHECK19-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 8
// CHECK19-NEXT: [[TMP90:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 6 // CHECK19-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to %struct.TT**
// CHECK19-NEXT: store i8* null, i8** [[TMP90]], align 4 // CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP90]], align 4
// CHECK19-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 7 // CHECK19-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 8
// CHECK19-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to double** // CHECK19-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to %struct.TT**
// CHECK19-NEXT: store double* [[VLA1]], double** [[TMP92]], align 4 // CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP92]], align 4
// CHECK19-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 7 // CHECK19-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 8
// CHECK19-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to double** // CHECK19-NEXT: store i8* null, i8** [[TMP93]], align 4
// CHECK19-NEXT: store double* [[VLA1]], double** [[TMP94]], align 4 // CHECK19-NEXT: [[TMP94:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK19-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK19-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK19-NEXT: store i64 [[TMP48]], i64* [[TMP95]], align 4 // CHECK19-NEXT: [[TMP96:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP94]], i8** [[TMP95]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: [[TMP96:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 7 // CHECK19-NEXT: [[TMP97:%.*]] = icmp ne i32 [[TMP96]], 0
// CHECK19-NEXT: store i8* null, i8** [[TMP96]], align 4 // CHECK19-NEXT: br i1 [[TMP97]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK19-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 8
// CHECK19-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to %struct.TT**
// CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP98]], align 4
// CHECK19-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 8
// CHECK19-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to %struct.TT**
// CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP100]], align 4
// CHECK19-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK19-NEXT: store i64 12, i64* [[TMP101]], align 4
// CHECK19-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 8
// CHECK19-NEXT: store i8* null, i8** [[TMP102]], align 4
// CHECK19-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK19-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK19-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK19-NEXT: [[TMP106:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP103]], i8** [[TMP104]], i64* [[TMP105]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: [[TMP107:%.*]] = icmp ne i32 [[TMP106]], 0
// CHECK19-NEXT: br i1 [[TMP107]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK19: omp_offload.failed16: // CHECK19: omp_offload.failed16:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT17]]
// CHECK19: omp_offload.cont17: // CHECK19: omp_offload.cont17:
// CHECK19-NEXT: br label [[OMP_IF_END19:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END19:%.*]]
// CHECK19: omp_if.else18: // CHECK19: omp_if.else18:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_IF_END19]] // CHECK19-NEXT: br label [[OMP_IF_END19]]
// CHECK19: omp_if.end19: // CHECK19: omp_if.end19:
// CHECK19-NEXT: [[TMP108:%.*]] = load i32, i32* [[A]], align 4 // CHECK19-NEXT: [[TMP98:%.*]] = load i32, i32* [[A]], align 4
// CHECK19-NEXT: [[TMP109:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: [[TMP99:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP109]]) // CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP99]])
// CHECK19-NEXT: ret i32 [[TMP108]] // CHECK19-NEXT: ret i32 [[TMP98]]
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100
// CHECK19-SAME: () #[[ATTR2:[0-9]+]] { // CHECK19-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
▲ Show 20 Lines • Show All 312 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK19-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK19-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK19-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK19-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK19-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK19-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 13 Lines
// CHECK19-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK19-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK19-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK19-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK19-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK19-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK19-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK19-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK19-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK19-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK19-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK19-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK19-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK19-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK19-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK19-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK19-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK19-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK19-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK19-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK19-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK19-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK19-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK19-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK19-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK19-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK19-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK19-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK19-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK19-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK19-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK19-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK19-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK19-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK19-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK19-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK19-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK19-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK19-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK19-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK19-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK19-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK19-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK19-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4 // CHECK19-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK19-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK19-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK19-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK19-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK19-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK19-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK19-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: br label [[OMP_IF_END:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK19: omp_if.else: // CHECK19: omp_if.else:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_IF_END]] // CHECK19-NEXT: br label [[OMP_IF_END]]
// CHECK19: omp_if.end: // CHECK19: omp_if.end:
// CHECK19-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK19-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK19-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK19-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK19-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK19-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK19-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK19-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK19-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK19-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK19-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK19-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK19-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK19-NEXT: ret i32 [[ADD3]] // CHECK19-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK19-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK19-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK19-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK19-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK19-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK19-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK19-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK19-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK19-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK19-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: br label [[OMP_IF_END:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK19: omp_if.else: // CHECK19: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK19-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK19-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK19-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK19-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK19-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK19-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: br label [[OMP_IF_END:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK19: omp_if.else: // CHECK19: omp_if.else:
Show All 20 Lines
// CHECK19-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK19-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK19-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK19-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK19-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 49 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK19-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK19-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK19-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK19-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK19-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK19-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK19-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK19-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK19-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK19-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK19-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK19-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
▲ Show 20 Lines • Show All 41 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK19-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK19-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK19-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK19-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK19-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK19-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
▲ Show 20 Lines • Show All 47 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[AA_CASTED3:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AA_CASTED3:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4
// CHECK20-NEXT: [[A_CASTED10:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A_CASTED10:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS13:%.*]] = alloca [9 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS13:%.*]] = alloca [9 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS14:%.*]] = alloca [9 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS14:%.*]] = alloca [9 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS15:%.*]] = alloca [9 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS15:%.*]] = alloca [9 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]]) // CHECK20-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
// CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK20-NEXT: store i32 0, i32* [[A]], align 4 // CHECK20-NEXT: store i32 0, i32* [[A]], align 4
// CHECK20-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK20-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK20-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK20-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK20-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK20-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 85 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[TMP47:%.*]] = mul nuw i32 [[TMP46]], 8 // CHECK20-NEXT: [[TMP47:%.*]] = mul nuw i32 [[TMP46]], 8
// CHECK20-NEXT: [[TMP48:%.*]] = sext i32 [[TMP47]] to i64 // CHECK20-NEXT: [[TMP48:%.*]] = sext i32 [[TMP47]] to i64
// CHECK20-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0 // CHECK20-NEXT: [[TMP49:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK20-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32* // CHECK20-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32*
// CHECK20-NEXT: store i32 [[TMP42]], i32* [[TMP50]], align 4 // CHECK20-NEXT: store i32 [[TMP42]], i32* [[TMP50]], align 4
// CHECK20-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0 // CHECK20-NEXT: [[TMP51:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK20-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32* // CHECK20-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32*
// CHECK20-NEXT: store i32 [[TMP42]], i32* [[TMP52]], align 4 // CHECK20-NEXT: store i32 [[TMP42]], i32* [[TMP52]], align 4
// CHECK20-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK20-NEXT: [[TMP53:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 0
// CHECK20-NEXT: store i64 4, i64* [[TMP53]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP53]], align 4
// CHECK20-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 0 // CHECK20-NEXT: [[TMP54:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 1
// CHECK20-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK20-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to [10 x float]**
// CHECK20-NEXT: [[TMP55:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 1 // CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP55]], align 4
// CHECK20-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to [10 x float]** // CHECK20-NEXT: [[TMP56:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 1
// CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP56]], align 4 // CHECK20-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to [10 x float]**
// CHECK20-NEXT: [[TMP57:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 1 // CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP57]], align 4
// CHECK20-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to [10 x float]** // CHECK20-NEXT: [[TMP58:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 1
// CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP58]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP58]], align 4
// CHECK20-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK20-NEXT: [[TMP59:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 2
// CHECK20-NEXT: store i64 40, i64* [[TMP59]], align 4 // CHECK20-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i32*
// CHECK20-NEXT: [[TMP60:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP60]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP60]], align 4 // CHECK20-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 2
// CHECK20-NEXT: [[TMP61:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 2
// CHECK20-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i32* // CHECK20-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i32*
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP62]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP62]], align 4
// CHECK20-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 2 // CHECK20-NEXT: [[TMP63:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 2
// CHECK20-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32* // CHECK20-NEXT: store i8* null, i8** [[TMP63]], align 4
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP64]], align 4 // CHECK20-NEXT: [[TMP64:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 3
// CHECK20-NEXT: [[TMP65:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK20-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to float**
// CHECK20-NEXT: store i64 4, i64* [[TMP65]], align 4 // CHECK20-NEXT: store float* [[VLA]], float** [[TMP65]], align 4
// CHECK20-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 2 // CHECK20-NEXT: [[TMP66:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 3
// CHECK20-NEXT: store i8* null, i8** [[TMP66]], align 4 // CHECK20-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to float**
// CHECK20-NEXT: [[TMP67:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 3 // CHECK20-NEXT: store float* [[VLA]], float** [[TMP67]], align 4
// CHECK20-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to float** // CHECK20-NEXT: store i64 [[TMP45]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK20-NEXT: store float* [[VLA]], float** [[TMP68]], align 4 // CHECK20-NEXT: [[TMP68:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 3
// CHECK20-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 3 // CHECK20-NEXT: store i8* null, i8** [[TMP68]], align 4
// CHECK20-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to float** // CHECK20-NEXT: [[TMP69:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 4
// CHECK20-NEXT: store float* [[VLA]], float** [[TMP70]], align 4 // CHECK20-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to [5 x [10 x double]]**
// CHECK20-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP70]], align 4
// CHECK20-NEXT: store i64 [[TMP45]], i64* [[TMP71]], align 4 // CHECK20-NEXT: [[TMP71:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 4
// CHECK20-NEXT: [[TMP72:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 3 // CHECK20-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [5 x [10 x double]]**
// CHECK20-NEXT: store i8* null, i8** [[TMP72]], align 4 // CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP72]], align 4
// CHECK20-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 4 // CHECK20-NEXT: [[TMP73:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 4
// CHECK20-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [5 x [10 x double]]** // CHECK20-NEXT: store i8* null, i8** [[TMP73]], align 4
// CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP74]], align 4 // CHECK20-NEXT: [[TMP74:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 5
// CHECK20-NEXT: [[TMP75:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 4 // CHECK20-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32*
// CHECK20-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [5 x [10 x double]]** // CHECK20-NEXT: store i32 5, i32* [[TMP75]], align 4
// CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP76]], align 4 // CHECK20-NEXT: [[TMP76:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 5
// CHECK20-NEXT: [[TMP77:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK20-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32*
// CHECK20-NEXT: store i64 400, i64* [[TMP77]], align 4 // CHECK20-NEXT: store i32 5, i32* [[TMP77]], align 4
// CHECK20-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 4 // CHECK20-NEXT: [[TMP78:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 5
// CHECK20-NEXT: store i8* null, i8** [[TMP78]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP78]], align 4
// CHECK20-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 5 // CHECK20-NEXT: [[TMP79:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 6
// CHECK20-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK20-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32*
// CHECK20-NEXT: store i32 5, i32* [[TMP80]], align 4 // CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP80]], align 4
// CHECK20-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 5 // CHECK20-NEXT: [[TMP81:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 6
// CHECK20-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32* // CHECK20-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32*
// CHECK20-NEXT: store i32 5, i32* [[TMP82]], align 4 // CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP82]], align 4
// CHECK20-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK20-NEXT: [[TMP83:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 6
// CHECK20-NEXT: store i64 4, i64* [[TMP83]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP83]], align 4
// CHECK20-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 5 // CHECK20-NEXT: [[TMP84:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 7
// CHECK20-NEXT: store i8* null, i8** [[TMP84]], align 4 // CHECK20-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to double**
// CHECK20-NEXT: [[TMP85:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 6 // CHECK20-NEXT: store double* [[VLA1]], double** [[TMP85]], align 4
// CHECK20-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32* // CHECK20-NEXT: [[TMP86:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 7
// CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP86]], align 4 // CHECK20-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to double**
// CHECK20-NEXT: [[TMP87:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 6 // CHECK20-NEXT: store double* [[VLA1]], double** [[TMP87]], align 4
// CHECK20-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32* // CHECK20-NEXT: store i64 [[TMP48]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP88]], align 4 // CHECK20-NEXT: [[TMP88:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 7
// CHECK20-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK20-NEXT: store i8* null, i8** [[TMP88]], align 4
// CHECK20-NEXT: store i64 4, i64* [[TMP89]], align 4 // CHECK20-NEXT: [[TMP89:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 8
// CHECK20-NEXT: [[TMP90:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 6 // CHECK20-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to %struct.TT**
// CHECK20-NEXT: store i8* null, i8** [[TMP90]], align 4 // CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP90]], align 4
// CHECK20-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 7 // CHECK20-NEXT: [[TMP91:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 8
// CHECK20-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to double** // CHECK20-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to %struct.TT**
// CHECK20-NEXT: store double* [[VLA1]], double** [[TMP92]], align 4 // CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP92]], align 4
// CHECK20-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 7 // CHECK20-NEXT: [[TMP93:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 8
// CHECK20-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to double** // CHECK20-NEXT: store i8* null, i8** [[TMP93]], align 4
// CHECK20-NEXT: store double* [[VLA1]], double** [[TMP94]], align 4 // CHECK20-NEXT: [[TMP94:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK20-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK20-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK20-NEXT: store i64 [[TMP48]], i64* [[TMP95]], align 4 // CHECK20-NEXT: [[TMP96:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP94]], i8** [[TMP95]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: [[TMP96:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 7 // CHECK20-NEXT: [[TMP97:%.*]] = icmp ne i32 [[TMP96]], 0
// CHECK20-NEXT: store i8* null, i8** [[TMP96]], align 4 // CHECK20-NEXT: br i1 [[TMP97]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK20-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 8
// CHECK20-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to %struct.TT**
// CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP98]], align 4
// CHECK20-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 8
// CHECK20-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to %struct.TT**
// CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP100]], align 4
// CHECK20-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK20-NEXT: store i64 12, i64* [[TMP101]], align 4
// CHECK20-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS15]], i32 0, i32 8
// CHECK20-NEXT: store i8* null, i8** [[TMP102]], align 4
// CHECK20-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK20-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK20-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK20-NEXT: [[TMP106:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144.region_id, i32 9, i8** [[TMP103]], i8** [[TMP104]], i64* [[TMP105]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: [[TMP107:%.*]] = icmp ne i32 [[TMP106]], 0
// CHECK20-NEXT: br i1 [[TMP107]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK20: omp_offload.failed16: // CHECK20: omp_offload.failed16:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT17]]
// CHECK20: omp_offload.cont17: // CHECK20: omp_offload.cont17:
// CHECK20-NEXT: br label [[OMP_IF_END19:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END19:%.*]]
// CHECK20: omp_if.else18: // CHECK20: omp_if.else18:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l144(i32 [[TMP42]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_IF_END19]] // CHECK20-NEXT: br label [[OMP_IF_END19]]
// CHECK20: omp_if.end19: // CHECK20: omp_if.end19:
// CHECK20-NEXT: [[TMP108:%.*]] = load i32, i32* [[A]], align 4 // CHECK20-NEXT: [[TMP98:%.*]] = load i32, i32* [[A]], align 4
// CHECK20-NEXT: [[TMP109:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: [[TMP99:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP109]]) // CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP99]])
// CHECK20-NEXT: ret i32 [[TMP108]] // CHECK20-NEXT: ret i32 [[TMP98]]
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100 // CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l100
// CHECK20-SAME: () #[[ATTR2:[0-9]+]] { // CHECK20-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
▲ Show 20 Lines • Show All 312 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK20-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK20-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK20-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK20-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK20-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK20-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK20-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK20-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK20-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 13 Lines
// CHECK20-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK20-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK20-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK20-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK20-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK20-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK20-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK20-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK20-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK20-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK20-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK20-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK20-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK20-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK20-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK20-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK20-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK20-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK20-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK20-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK20-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK20-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK20-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK20-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK20-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK20-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK20-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK20-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK20-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK20-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK20-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK20-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK20-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK20-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK20-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK20-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK20-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK20-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK20-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK20-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK20-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK20-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK20-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK20-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK20-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK20-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK20-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK20-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK20-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK20-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK20-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK20-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK20-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK20-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK20-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK20-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK20-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4 // CHECK20-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK20-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK20-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK20-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK20-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK20-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK20-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK20-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: br label [[OMP_IF_END:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK20: omp_if.else: // CHECK20: omp_if.else:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_IF_END]] // CHECK20-NEXT: br label [[OMP_IF_END]]
// CHECK20: omp_if.end: // CHECK20: omp_if.end:
// CHECK20-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK20-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK20-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK20-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK20-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK20-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK20-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK20-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK20-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK20-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK20-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK20-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK20-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK20-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK20-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK20-NEXT: ret i32 [[ADD3]] // CHECK20-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK20-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK20-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK20-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK20-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK20-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK20-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK20-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK20-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK20-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK20-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK20-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK20-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l198(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: br label [[OMP_IF_END:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK20: omp_if.else: // CHECK20: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK20-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK20-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK20-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK20-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK20-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK20-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: br label [[OMP_IF_END:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK20: omp_if.else: // CHECK20: omp_if.else:
Show All 20 Lines
// CHECK20-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK20-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK20-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK20-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK20-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK20-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK20-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK20-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK20-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK20-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 49 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK20-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK20-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK20-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK20-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK20-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK20-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK20-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK20-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK20-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK20-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK20-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK20-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK20-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK20-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
▲ Show 20 Lines • Show All 41 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK20-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK20-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK20-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK20-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK20-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK20-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK20-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK20-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK20-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
▲ Show 20 Lines • Show All 1,653 Lines • Show Last 20 Lines

clang/test/OpenMP/target_parallel_for_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 350 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 0, i32* [[A]], align 4 // CHECK1-NEXT: store i32 0, i32* [[A]], align 4
// CHECK1-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK1-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK1-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK1-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 153 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP87:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK1-NEXT: [[TMP87:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK1-NEXT: [[TMP88:%.*]] = mul nuw i64 [[TMP87]], 8 // CHECK1-NEXT: [[TMP88:%.*]] = mul nuw i64 [[TMP87]], 8
// CHECK1-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK1-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK1-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK1-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64*
// CHECK1-NEXT: store i64 [[TMP82]], i64* [[TMP90]], align 8 // CHECK1-NEXT: store i64 [[TMP82]], i64* [[TMP90]], align 8
// CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK1-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i64* // CHECK1-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i64*
// CHECK1-NEXT: store i64 [[TMP82]], i64* [[TMP92]], align 8 // CHECK1-NEXT: store i64 [[TMP82]], i64* [[TMP92]], align 8
// CHECK1-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK1-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK1-NEXT: store i64 4, i64* [[TMP93]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP93]], align 8
// CHECK1-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK1-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK1-NEXT: store i8* null, i8** [[TMP94]], align 8 // CHECK1-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [10 x float]**
// CHECK1-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP95]], align 8
// CHECK1-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to [10 x float]** // CHECK1-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP96]], align 8 // CHECK1-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to [10 x float]**
// CHECK1-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP97]], align 8
// CHECK1-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to [10 x float]** // CHECK1-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP98]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP98]], align 8
// CHECK1-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK1-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK1-NEXT: store i64 40, i64* [[TMP99]], align 8 // CHECK1-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i64*
// CHECK1-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP100]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP100]], align 8 // CHECK1-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK1-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK1-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i64* // CHECK1-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i64*
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP102]], align 8 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP102]], align 8
// CHECK1-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK1-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK1-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to i64* // CHECK1-NEXT: store i8* null, i8** [[TMP103]], align 8
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP104]], align 8 // CHECK1-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK1-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK1-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to float**
// CHECK1-NEXT: store i64 8, i64* [[TMP105]], align 8 // CHECK1-NEXT: store float* [[VLA]], float** [[TMP105]], align 8
// CHECK1-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK1-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK1-NEXT: store i8* null, i8** [[TMP106]], align 8 // CHECK1-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to float**
// CHECK1-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3 // CHECK1-NEXT: store float* [[VLA]], float** [[TMP107]], align 8
// CHECK1-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to float** // CHECK1-NEXT: store i64 [[TMP86]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK1-NEXT: store float* [[VLA]], float** [[TMP108]], align 8 // CHECK1-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK1-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3 // CHECK1-NEXT: store i8* null, i8** [[TMP108]], align 8
// CHECK1-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to float** // CHECK1-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4
// CHECK1-NEXT: store float* [[VLA]], float** [[TMP110]], align 8 // CHECK1-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to [5 x [10 x double]]**
// CHECK1-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP110]], align 8
// CHECK1-NEXT: store i64 [[TMP86]], i64* [[TMP111]], align 8 // CHECK1-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4
// CHECK1-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3 // CHECK1-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to [5 x [10 x double]]**
// CHECK1-NEXT: store i8* null, i8** [[TMP112]], align 8 // CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP112]], align 8
// CHECK1-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4 // CHECK1-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4
// CHECK1-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to [5 x [10 x double]]** // CHECK1-NEXT: store i8* null, i8** [[TMP113]], align 8
// CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP114]], align 8 // CHECK1-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5
// CHECK1-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4 // CHECK1-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to i64*
// CHECK1-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]** // CHECK1-NEXT: store i64 5, i64* [[TMP115]], align 8
// CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 8 // CHECK1-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5
// CHECK1-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK1-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to i64*
// CHECK1-NEXT: store i64 400, i64* [[TMP117]], align 8 // CHECK1-NEXT: store i64 5, i64* [[TMP117]], align 8
// CHECK1-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4 // CHECK1-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5
// CHECK1-NEXT: store i8* null, i8** [[TMP118]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP118]], align 8
// CHECK1-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5 // CHECK1-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6
// CHECK1-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i64* // CHECK1-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i64*
// CHECK1-NEXT: store i64 5, i64* [[TMP120]], align 8 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP120]], align 8
// CHECK1-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5 // CHECK1-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6
// CHECK1-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to i64* // CHECK1-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to i64*
// CHECK1-NEXT: store i64 5, i64* [[TMP122]], align 8 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP122]], align 8
// CHECK1-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK1-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6
// CHECK1-NEXT: store i64 8, i64* [[TMP123]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP123]], align 8
// CHECK1-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5 // CHECK1-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7
// CHECK1-NEXT: store i8* null, i8** [[TMP124]], align 8 // CHECK1-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to double**
// CHECK1-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6 // CHECK1-NEXT: store double* [[VLA1]], double** [[TMP125]], align 8
// CHECK1-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64* // CHECK1-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP126]], align 8 // CHECK1-NEXT: [[TMP127:%.*]] = bitcast i8** [[TMP126]] to double**
// CHECK1-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6 // CHECK1-NEXT: store double* [[VLA1]], double** [[TMP127]], align 8
// CHECK1-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64* // CHECK1-NEXT: store i64 [[TMP88]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP128]], align 8 // CHECK1-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7
// CHECK1-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK1-NEXT: store i8* null, i8** [[TMP128]], align 8
// CHECK1-NEXT: store i64 8, i64* [[TMP129]], align 8 // CHECK1-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8
// CHECK1-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6 // CHECK1-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to %struct.TT**
// CHECK1-NEXT: store i8* null, i8** [[TMP130]], align 8 // CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP130]], align 8
// CHECK1-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7 // CHECK1-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8
// CHECK1-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to double** // CHECK1-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to %struct.TT**
// CHECK1-NEXT: store double* [[VLA1]], double** [[TMP132]], align 8 // CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP132]], align 8
// CHECK1-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7 // CHECK1-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK1-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double** // CHECK1-NEXT: store i8* null, i8** [[TMP133]], align 8
// CHECK1-NEXT: store double* [[VLA1]], double** [[TMP134]], align 8 // CHECK1-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK1-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK1-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i64*
// CHECK1-NEXT: store i64 [[TMP88]], i64* [[TMP135]], align 8 // CHECK1-NEXT: store i64 [[TMP84]], i64* [[TMP135]], align 8
// CHECK1-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7 // CHECK1-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK1-NEXT: store i8* null, i8** [[TMP136]], align 8 // CHECK1-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i64*
// CHECK1-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8 // CHECK1-NEXT: store i64 [[TMP84]], i64* [[TMP137]], align 8
// CHECK1-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT** // CHECK1-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP138]], align 8
// CHECK1-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8 // CHECK1-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK1-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to %struct.TT** // CHECK1-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP140]], align 8 // CHECK1-NEXT: [[TMP141:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP139]], i8** [[TMP140]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP141:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK1-NEXT: [[TMP142:%.*]] = icmp ne i32 [[TMP141]], 0
// CHECK1-NEXT: store i64 16, i64* [[TMP141]], align 8 // CHECK1-NEXT: br i1 [[TMP142]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK1-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK1-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK1-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK1-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i64*
// CHECK1-NEXT: store i64 [[TMP84]], i64* [[TMP144]], align 8
// CHECK1-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK1-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i64*
// CHECK1-NEXT: store i64 [[TMP84]], i64* [[TMP146]], align 8
// CHECK1-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK1-NEXT: store i64 4, i64* [[TMP147]], align 8
// CHECK1-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK1-NEXT: store i8* null, i8** [[TMP148]], align 8
// CHECK1-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK1-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK1-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK1-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0
// CHECK1-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK1: omp_offload.failed23: // CHECK1: omp_offload.failed23:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT24]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT24]]
// CHECK1: omp_offload.cont24: // CHECK1: omp_offload.cont24:
// CHECK1-NEXT: br label [[OMP_IF_END26:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END26:%.*]]
// CHECK1: omp_if.else25: // CHECK1: omp_if.else25:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_IF_END26]] // CHECK1-NEXT: br label [[OMP_IF_END26]]
// CHECK1: omp_if.end26: // CHECK1: omp_if.end26:
// CHECK1-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK1-NEXT: [[TMP143:%.*]] = load i32, i32* [[A]], align 4
// CHECK1-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP144:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP144]])
// CHECK1-NEXT: ret i32 [[TMP154]] // CHECK1-NEXT: ret i32 [[TMP143]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK1-SAME: () #[[ATTR2:[0-9]+]] { // CHECK1-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
▲ Show 20 Lines • Show All 756 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK1-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK1-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK1-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK1-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 14 Lines
// CHECK1-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK1-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK1-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK1-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK1-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK1-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK1-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK1-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK1-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK1-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK1-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK1-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK1-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK1-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK1-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK1-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK1-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK1-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK1-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK1-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK1-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK1-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8 // CHECK1-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK1-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK1-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK1-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_IF_END]] // CHECK1-NEXT: br label [[OMP_IF_END]]
// CHECK1: omp_if.end: // CHECK1: omp_if.end:
// CHECK1-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK1-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK1-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK1-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK1-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK1-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK1-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK1-NEXT: ret i32 [[ADD4]] // CHECK1-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK1-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK1-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK1-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK1-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK1-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
Show All 22 Lines
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 102 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK1-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK1-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK1-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK1-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK1-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK1-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK1-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
Show All 29 Lines
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 111 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 0, i32* [[A]], align 4 // CHECK2-NEXT: store i32 0, i32* [[A]], align 4
// CHECK2-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK2-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK2-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK2-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 153 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP87:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK2-NEXT: [[TMP87:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK2-NEXT: [[TMP88:%.*]] = mul nuw i64 [[TMP87]], 8 // CHECK2-NEXT: [[TMP88:%.*]] = mul nuw i64 [[TMP87]], 8
// CHECK2-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK2-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK2-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK2-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64*
// CHECK2-NEXT: store i64 [[TMP82]], i64* [[TMP90]], align 8 // CHECK2-NEXT: store i64 [[TMP82]], i64* [[TMP90]], align 8
// CHECK2-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK2-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK2-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i64* // CHECK2-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i64*
// CHECK2-NEXT: store i64 [[TMP82]], i64* [[TMP92]], align 8 // CHECK2-NEXT: store i64 [[TMP82]], i64* [[TMP92]], align 8
// CHECK2-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK2-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK2-NEXT: store i64 4, i64* [[TMP93]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP93]], align 8
// CHECK2-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK2-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP94]], align 8 // CHECK2-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [10 x float]**
// CHECK2-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP95]], align 8
// CHECK2-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to [10 x float]** // CHECK2-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP96]], align 8 // CHECK2-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to [10 x float]**
// CHECK2-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP97]], align 8
// CHECK2-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to [10 x float]** // CHECK2-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP98]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP98]], align 8
// CHECK2-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK2-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK2-NEXT: store i64 40, i64* [[TMP99]], align 8 // CHECK2-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i64*
// CHECK2-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP100]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP100]], align 8 // CHECK2-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK2-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK2-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i64* // CHECK2-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i64*
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP102]], align 8 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP102]], align 8
// CHECK2-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK2-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK2-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to i64* // CHECK2-NEXT: store i8* null, i8** [[TMP103]], align 8
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP104]], align 8 // CHECK2-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK2-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK2-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to float**
// CHECK2-NEXT: store i64 8, i64* [[TMP105]], align 8 // CHECK2-NEXT: store float* [[VLA]], float** [[TMP105]], align 8
// CHECK2-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK2-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK2-NEXT: store i8* null, i8** [[TMP106]], align 8 // CHECK2-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to float**
// CHECK2-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3 // CHECK2-NEXT: store float* [[VLA]], float** [[TMP107]], align 8
// CHECK2-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to float** // CHECK2-NEXT: store i64 [[TMP86]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK2-NEXT: store float* [[VLA]], float** [[TMP108]], align 8 // CHECK2-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK2-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3 // CHECK2-NEXT: store i8* null, i8** [[TMP108]], align 8
// CHECK2-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to float** // CHECK2-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4
// CHECK2-NEXT: store float* [[VLA]], float** [[TMP110]], align 8 // CHECK2-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to [5 x [10 x double]]**
// CHECK2-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP110]], align 8
// CHECK2-NEXT: store i64 [[TMP86]], i64* [[TMP111]], align 8 // CHECK2-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4
// CHECK2-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3 // CHECK2-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to [5 x [10 x double]]**
// CHECK2-NEXT: store i8* null, i8** [[TMP112]], align 8 // CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP112]], align 8
// CHECK2-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4 // CHECK2-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4
// CHECK2-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to [5 x [10 x double]]** // CHECK2-NEXT: store i8* null, i8** [[TMP113]], align 8
// CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP114]], align 8 // CHECK2-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5
// CHECK2-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4 // CHECK2-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to i64*
// CHECK2-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]** // CHECK2-NEXT: store i64 5, i64* [[TMP115]], align 8
// CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 8 // CHECK2-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5
// CHECK2-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK2-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to i64*
// CHECK2-NEXT: store i64 400, i64* [[TMP117]], align 8 // CHECK2-NEXT: store i64 5, i64* [[TMP117]], align 8
// CHECK2-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4 // CHECK2-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5
// CHECK2-NEXT: store i8* null, i8** [[TMP118]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP118]], align 8
// CHECK2-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5 // CHECK2-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6
// CHECK2-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i64* // CHECK2-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i64*
// CHECK2-NEXT: store i64 5, i64* [[TMP120]], align 8 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP120]], align 8
// CHECK2-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5 // CHECK2-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6
// CHECK2-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to i64* // CHECK2-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to i64*
// CHECK2-NEXT: store i64 5, i64* [[TMP122]], align 8 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP122]], align 8
// CHECK2-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK2-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6
// CHECK2-NEXT: store i64 8, i64* [[TMP123]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP123]], align 8
// CHECK2-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5 // CHECK2-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7
// CHECK2-NEXT: store i8* null, i8** [[TMP124]], align 8 // CHECK2-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to double**
// CHECK2-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6 // CHECK2-NEXT: store double* [[VLA1]], double** [[TMP125]], align 8
// CHECK2-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64* // CHECK2-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7
// CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP126]], align 8 // CHECK2-NEXT: [[TMP127:%.*]] = bitcast i8** [[TMP126]] to double**
// CHECK2-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6 // CHECK2-NEXT: store double* [[VLA1]], double** [[TMP127]], align 8
// CHECK2-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64* // CHECK2-NEXT: store i64 [[TMP88]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP128]], align 8 // CHECK2-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7
// CHECK2-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK2-NEXT: store i8* null, i8** [[TMP128]], align 8
// CHECK2-NEXT: store i64 8, i64* [[TMP129]], align 8 // CHECK2-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8
// CHECK2-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6 // CHECK2-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to %struct.TT**
// CHECK2-NEXT: store i8* null, i8** [[TMP130]], align 8 // CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP130]], align 8
// CHECK2-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7 // CHECK2-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8
// CHECK2-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to double** // CHECK2-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to %struct.TT**
// CHECK2-NEXT: store double* [[VLA1]], double** [[TMP132]], align 8 // CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP132]], align 8
// CHECK2-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7 // CHECK2-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK2-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double** // CHECK2-NEXT: store i8* null, i8** [[TMP133]], align 8
// CHECK2-NEXT: store double* [[VLA1]], double** [[TMP134]], align 8 // CHECK2-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK2-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK2-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i64*
// CHECK2-NEXT: store i64 [[TMP88]], i64* [[TMP135]], align 8 // CHECK2-NEXT: store i64 [[TMP84]], i64* [[TMP135]], align 8
// CHECK2-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7 // CHECK2-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK2-NEXT: store i8* null, i8** [[TMP136]], align 8 // CHECK2-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i64*
// CHECK2-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8 // CHECK2-NEXT: store i64 [[TMP84]], i64* [[TMP137]], align 8
// CHECK2-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT** // CHECK2-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP138]], align 8
// CHECK2-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8 // CHECK2-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK2-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to %struct.TT** // CHECK2-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP140]], align 8 // CHECK2-NEXT: [[TMP141:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP139]], i8** [[TMP140]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP141:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK2-NEXT: [[TMP142:%.*]] = icmp ne i32 [[TMP141]], 0
// CHECK2-NEXT: store i64 16, i64* [[TMP141]], align 8 // CHECK2-NEXT: br i1 [[TMP142]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK2-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK2-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK2-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK2-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i64*
// CHECK2-NEXT: store i64 [[TMP84]], i64* [[TMP144]], align 8
// CHECK2-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK2-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i64*
// CHECK2-NEXT: store i64 [[TMP84]], i64* [[TMP146]], align 8
// CHECK2-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK2-NEXT: store i64 4, i64* [[TMP147]], align 8
// CHECK2-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK2-NEXT: store i8* null, i8** [[TMP148]], align 8
// CHECK2-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK2-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK2-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK2-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0
// CHECK2-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK2: omp_offload.failed23: // CHECK2: omp_offload.failed23:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT24]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT24]]
// CHECK2: omp_offload.cont24: // CHECK2: omp_offload.cont24:
// CHECK2-NEXT: br label [[OMP_IF_END26:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END26:%.*]]
// CHECK2: omp_if.else25: // CHECK2: omp_if.else25:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_IF_END26]] // CHECK2-NEXT: br label [[OMP_IF_END26]]
// CHECK2: omp_if.end26: // CHECK2: omp_if.end26:
// CHECK2-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK2-NEXT: [[TMP143:%.*]] = load i32, i32* [[A]], align 4
// CHECK2-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: [[TMP144:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP144]])
// CHECK2-NEXT: ret i32 [[TMP154]] // CHECK2-NEXT: ret i32 [[TMP143]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK2-SAME: () #[[ATTR2:[0-9]+]] { // CHECK2-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
▲ Show 20 Lines • Show All 756 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK2-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK2-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK2-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK2-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 14 Lines
// CHECK2-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK2-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK2-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK2-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK2-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK2-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK2-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK2-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK2-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK2-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK2-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK2-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK2-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK2-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK2-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK2-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK2-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK2-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK2-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK2-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK2-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK2-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK2-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK2-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK2-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK2-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK2-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK2-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK2-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK2-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8 // CHECK2-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK2-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK2-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK2-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK2-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK2-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK2-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK2-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_IF_END]] // CHECK2-NEXT: br label [[OMP_IF_END]]
// CHECK2: omp_if.end: // CHECK2: omp_if.end:
// CHECK2-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK2-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK2-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK2-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK2-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK2-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK2-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK2-NEXT: ret i32 [[ADD4]] // CHECK2-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK2-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK2-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK2-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK2-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK2-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK2-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK2-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK2-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK2-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
Show All 22 Lines
// CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 102 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK2-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK2-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK2-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK2-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK2-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK2-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK2-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
Show All 29 Lines
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[A]], align 4 // CHECK3-NEXT: store i32 0, i32* [[A]], align 4
// CHECK3-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK3-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK3-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK3-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 144 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP85:%.*]] = mul nuw i32 [[TMP84]], 8 // CHECK3-NEXT: [[TMP85:%.*]] = mul nuw i32 [[TMP84]], 8
// CHECK3-NEXT: [[TMP86:%.*]] = sext i32 [[TMP85]] to i64 // CHECK3-NEXT: [[TMP86:%.*]] = sext i32 [[TMP85]] to i64
// CHECK3-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0 // CHECK3-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK3-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32* // CHECK3-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32*
// CHECK3-NEXT: store i32 [[TMP78]], i32* [[TMP88]], align 4 // CHECK3-NEXT: store i32 [[TMP78]], i32* [[TMP88]], align 4
// CHECK3-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0 // CHECK3-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK3-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK3-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32*
// CHECK3-NEXT: store i32 [[TMP78]], i32* [[TMP90]], align 4 // CHECK3-NEXT: store i32 [[TMP78]], i32* [[TMP90]], align 4
// CHECK3-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK3-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0
// CHECK3-NEXT: store i64 4, i64* [[TMP91]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP91]], align 4
// CHECK3-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0 // CHECK3-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP92]], align 4 // CHECK3-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [10 x float]**
// CHECK3-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1 // CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP93]], align 4
// CHECK3-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to [10 x float]** // CHECK3-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1
// CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP94]], align 4 // CHECK3-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [10 x float]**
// CHECK3-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1 // CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP95]], align 4
// CHECK3-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to [10 x float]** // CHECK3-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1
// CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP96]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP96]], align 4
// CHECK3-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK3-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK3-NEXT: store i64 40, i64* [[TMP97]], align 4 // CHECK3-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32*
// CHECK3-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP98]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP98]], align 4 // CHECK3-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2
// CHECK3-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK3-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i32* // CHECK3-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP100]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP100]], align 4
// CHECK3-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2 // CHECK3-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2
// CHECK3-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i32* // CHECK3-NEXT: store i8* null, i8** [[TMP101]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP102]], align 4 // CHECK3-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3
// CHECK3-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK3-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to float**
// CHECK3-NEXT: store i64 4, i64* [[TMP103]], align 4 // CHECK3-NEXT: store float* [[VLA]], float** [[TMP103]], align 4
// CHECK3-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2 // CHECK3-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP104]], align 4 // CHECK3-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to float**
// CHECK3-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3 // CHECK3-NEXT: store float* [[VLA]], float** [[TMP105]], align 4
// CHECK3-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to float** // CHECK3-NEXT: store i64 [[TMP83]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK3-NEXT: store float* [[VLA]], float** [[TMP106]], align 4 // CHECK3-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3
// CHECK3-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3 // CHECK3-NEXT: store i8* null, i8** [[TMP106]], align 4
// CHECK3-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to float** // CHECK3-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4
// CHECK3-NEXT: store float* [[VLA]], float** [[TMP108]], align 4 // CHECK3-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to [5 x [10 x double]]**
// CHECK3-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP108]], align 4
// CHECK3-NEXT: store i64 [[TMP83]], i64* [[TMP109]], align 4 // CHECK3-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4
// CHECK3-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3 // CHECK3-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to [5 x [10 x double]]**
// CHECK3-NEXT: store i8* null, i8** [[TMP110]], align 4 // CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP110]], align 4
// CHECK3-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4 // CHECK3-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4
// CHECK3-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to [5 x [10 x double]]** // CHECK3-NEXT: store i8* null, i8** [[TMP111]], align 4
// CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP112]], align 4 // CHECK3-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5
// CHECK3-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4 // CHECK3-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i32*
// CHECK3-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to [5 x [10 x double]]** // CHECK3-NEXT: store i32 5, i32* [[TMP113]], align 4
// CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP114]], align 4 // CHECK3-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5
// CHECK3-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK3-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to i32*
// CHECK3-NEXT: store i64 400, i64* [[TMP115]], align 4 // CHECK3-NEXT: store i32 5, i32* [[TMP115]], align 4
// CHECK3-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4 // CHECK3-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5
// CHECK3-NEXT: store i8* null, i8** [[TMP116]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP116]], align 4
// CHECK3-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5 // CHECK3-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6
// CHECK3-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to i32* // CHECK3-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to i32*
// CHECK3-NEXT: store i32 5, i32* [[TMP118]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP118]], align 4
// CHECK3-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5 // CHECK3-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6
// CHECK3-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i32* // CHECK3-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i32*
// CHECK3-NEXT: store i32 5, i32* [[TMP120]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP120]], align 4
// CHECK3-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK3-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6
// CHECK3-NEXT: store i64 4, i64* [[TMP121]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP121]], align 4
// CHECK3-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5 // CHECK3-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7
// CHECK3-NEXT: store i8* null, i8** [[TMP122]], align 4 // CHECK3-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to double**
// CHECK3-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6 // CHECK3-NEXT: store double* [[VLA1]], double** [[TMP123]], align 4
// CHECK3-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i32* // CHECK3-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7
// CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP124]], align 4 // CHECK3-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to double**
// CHECK3-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6 // CHECK3-NEXT: store double* [[VLA1]], double** [[TMP125]], align 4
// CHECK3-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32* // CHECK3-NEXT: store i64 [[TMP86]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP126]], align 4 // CHECK3-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7
// CHECK3-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK3-NEXT: store i8* null, i8** [[TMP126]], align 4
// CHECK3-NEXT: store i64 4, i64* [[TMP127]], align 4 // CHECK3-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8
// CHECK3-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6 // CHECK3-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to %struct.TT**
// CHECK3-NEXT: store i8* null, i8** [[TMP128]], align 4 // CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP128]], align 4
// CHECK3-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7 // CHECK3-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8
// CHECK3-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to double** // CHECK3-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to %struct.TT**
// CHECK3-NEXT: store double* [[VLA1]], double** [[TMP130]], align 4 // CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP130]], align 4
// CHECK3-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7 // CHECK3-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK3-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to double** // CHECK3-NEXT: store i8* null, i8** [[TMP131]], align 4
// CHECK3-NEXT: store double* [[VLA1]], double** [[TMP132]], align 4 // CHECK3-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK3-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK3-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to i32*
// CHECK3-NEXT: store i64 [[TMP86]], i64* [[TMP133]], align 4 // CHECK3-NEXT: store i32 [[TMP80]], i32* [[TMP133]], align 4
// CHECK3-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7 // CHECK3-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK3-NEXT: store i8* null, i8** [[TMP134]], align 4 // CHECK3-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i32*
// CHECK3-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8 // CHECK3-NEXT: store i32 [[TMP80]], i32* [[TMP135]], align 4
// CHECK3-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT** // CHECK3-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP136]], align 4
// CHECK3-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8 // CHECK3-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK3-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT** // CHECK3-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 4 // CHECK3-NEXT: [[TMP139:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP137]], i8** [[TMP138]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK3-NEXT: [[TMP140:%.*]] = icmp ne i32 [[TMP139]], 0
// CHECK3-NEXT: store i64 12, i64* [[TMP139]], align 4 // CHECK3-NEXT: br i1 [[TMP140]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK3-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK3-NEXT: store i8* null, i8** [[TMP140]], align 4
// CHECK3-NEXT: [[TMP141:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK3-NEXT: [[TMP142:%.*]] = bitcast i8** [[TMP141]] to i32*
// CHECK3-NEXT: store i32 [[TMP80]], i32* [[TMP142]], align 4
// CHECK3-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK3-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i32*
// CHECK3-NEXT: store i32 [[TMP80]], i32* [[TMP144]], align 4
// CHECK3-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK3-NEXT: store i64 4, i64* [[TMP145]], align 4
// CHECK3-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK3-NEXT: store i8* null, i8** [[TMP146]], align 4
// CHECK3-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK3-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK3-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK3-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP147]], i8** [[TMP148]], i64* [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK3-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK3: omp_offload.failed17: // CHECK3: omp_offload.failed17:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT18]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK3: omp_offload.cont18: // CHECK3: omp_offload.cont18:
// CHECK3-NEXT: br label [[OMP_IF_END20:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END20:%.*]]
// CHECK3: omp_if.else19: // CHECK3: omp_if.else19:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_IF_END20]] // CHECK3-NEXT: br label [[OMP_IF_END20]]
// CHECK3: omp_if.end20: // CHECK3: omp_if.end20:
// CHECK3-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4 // CHECK3-NEXT: [[TMP141:%.*]] = load i32, i32* [[A]], align 4
// CHECK3-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: [[TMP142:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP153]]) // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP142]])
// CHECK3-NEXT: ret i32 [[TMP152]] // CHECK3-NEXT: ret i32 [[TMP141]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK3-SAME: () #[[ATTR2:[0-9]+]] { // CHECK3-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
▲ Show 20 Lines • Show All 734 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK3-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK3-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK3-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 13 Lines
// CHECK3-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK3-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK3-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK3-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK3-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK3-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK3-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK3-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK3-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK3-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK3-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK3-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK3-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK3-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK3-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK3-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK3-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK3-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK3-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK3-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK3-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK3-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK3-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4 // CHECK3-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK3-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK3-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK3-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK3-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_IF_END]] // CHECK3-NEXT: br label [[OMP_IF_END]]
// CHECK3: omp_if.end: // CHECK3: omp_if.end:
// CHECK3-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK3-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK3-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK3-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK3-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK3-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK3-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK3-NEXT: ret i32 [[ADD3]] // CHECK3-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK3-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK3-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK3-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK3-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK3-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
Show All 20 Lines
// CHECK3-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK3-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 99 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK3-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK3-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK3-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK3-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK3-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK3-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK3-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
Show All 26 Lines
// CHECK3-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK3-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK3-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 109 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK4-NEXT: store i32 0, i32* [[A]], align 4 // CHECK4-NEXT: store i32 0, i32* [[A]], align 4
// CHECK4-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK4-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK4-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK4-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 144 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP85:%.*]] = mul nuw i32 [[TMP84]], 8 // CHECK4-NEXT: [[TMP85:%.*]] = mul nuw i32 [[TMP84]], 8
// CHECK4-NEXT: [[TMP86:%.*]] = sext i32 [[TMP85]] to i64 // CHECK4-NEXT: [[TMP86:%.*]] = sext i32 [[TMP85]] to i64
// CHECK4-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0 // CHECK4-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK4-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32* // CHECK4-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32*
// CHECK4-NEXT: store i32 [[TMP78]], i32* [[TMP88]], align 4 // CHECK4-NEXT: store i32 [[TMP78]], i32* [[TMP88]], align 4
// CHECK4-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0 // CHECK4-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK4-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK4-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32*
// CHECK4-NEXT: store i32 [[TMP78]], i32* [[TMP90]], align 4 // CHECK4-NEXT: store i32 [[TMP78]], i32* [[TMP90]], align 4
// CHECK4-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK4-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0
// CHECK4-NEXT: store i64 4, i64* [[TMP91]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP91]], align 4
// CHECK4-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0 // CHECK4-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1
// CHECK4-NEXT: store i8* null, i8** [[TMP92]], align 4 // CHECK4-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [10 x float]**
// CHECK4-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1 // CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP93]], align 4
// CHECK4-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to [10 x float]** // CHECK4-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1
// CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP94]], align 4 // CHECK4-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [10 x float]**
// CHECK4-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1 // CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP95]], align 4
// CHECK4-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to [10 x float]** // CHECK4-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1
// CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP96]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP96]], align 4
// CHECK4-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK4-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK4-NEXT: store i64 40, i64* [[TMP97]], align 4 // CHECK4-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32*
// CHECK4-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP98]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP98]], align 4 // CHECK4-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2
// CHECK4-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK4-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i32* // CHECK4-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i32*
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP100]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP100]], align 4
// CHECK4-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2 // CHECK4-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2
// CHECK4-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i32* // CHECK4-NEXT: store i8* null, i8** [[TMP101]], align 4
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP102]], align 4 // CHECK4-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3
// CHECK4-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK4-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to float**
// CHECK4-NEXT: store i64 4, i64* [[TMP103]], align 4 // CHECK4-NEXT: store float* [[VLA]], float** [[TMP103]], align 4
// CHECK4-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2 // CHECK4-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP104]], align 4 // CHECK4-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to float**
// CHECK4-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3 // CHECK4-NEXT: store float* [[VLA]], float** [[TMP105]], align 4
// CHECK4-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to float** // CHECK4-NEXT: store i64 [[TMP83]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK4-NEXT: store float* [[VLA]], float** [[TMP106]], align 4 // CHECK4-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3
// CHECK4-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3 // CHECK4-NEXT: store i8* null, i8** [[TMP106]], align 4
// CHECK4-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to float** // CHECK4-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4
// CHECK4-NEXT: store float* [[VLA]], float** [[TMP108]], align 4 // CHECK4-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to [5 x [10 x double]]**
// CHECK4-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP108]], align 4
// CHECK4-NEXT: store i64 [[TMP83]], i64* [[TMP109]], align 4 // CHECK4-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4
// CHECK4-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3 // CHECK4-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to [5 x [10 x double]]**
// CHECK4-NEXT: store i8* null, i8** [[TMP110]], align 4 // CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP110]], align 4
// CHECK4-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4 // CHECK4-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4
// CHECK4-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to [5 x [10 x double]]** // CHECK4-NEXT: store i8* null, i8** [[TMP111]], align 4
// CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP112]], align 4 // CHECK4-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5
// CHECK4-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4 // CHECK4-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i32*
// CHECK4-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to [5 x [10 x double]]** // CHECK4-NEXT: store i32 5, i32* [[TMP113]], align 4
// CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP114]], align 4 // CHECK4-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5
// CHECK4-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK4-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to i32*
// CHECK4-NEXT: store i64 400, i64* [[TMP115]], align 4 // CHECK4-NEXT: store i32 5, i32* [[TMP115]], align 4
// CHECK4-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4 // CHECK4-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5
// CHECK4-NEXT: store i8* null, i8** [[TMP116]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP116]], align 4
// CHECK4-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5 // CHECK4-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6
// CHECK4-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to i32* // CHECK4-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to i32*
// CHECK4-NEXT: store i32 5, i32* [[TMP118]], align 4 // CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP118]], align 4
// CHECK4-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5 // CHECK4-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6
// CHECK4-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i32* // CHECK4-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i32*
// CHECK4-NEXT: store i32 5, i32* [[TMP120]], align 4 // CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP120]], align 4
// CHECK4-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK4-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6
// CHECK4-NEXT: store i64 4, i64* [[TMP121]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP121]], align 4
// CHECK4-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5 // CHECK4-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7
// CHECK4-NEXT: store i8* null, i8** [[TMP122]], align 4 // CHECK4-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to double**
// CHECK4-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6 // CHECK4-NEXT: store double* [[VLA1]], double** [[TMP123]], align 4
// CHECK4-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i32* // CHECK4-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7
// CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP124]], align 4 // CHECK4-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to double**
// CHECK4-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6 // CHECK4-NEXT: store double* [[VLA1]], double** [[TMP125]], align 4
// CHECK4-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32* // CHECK4-NEXT: store i64 [[TMP86]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP126]], align 4 // CHECK4-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7
// CHECK4-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK4-NEXT: store i8* null, i8** [[TMP126]], align 4
// CHECK4-NEXT: store i64 4, i64* [[TMP127]], align 4 // CHECK4-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8
// CHECK4-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6 // CHECK4-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to %struct.TT**
// CHECK4-NEXT: store i8* null, i8** [[TMP128]], align 4 // CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP128]], align 4
// CHECK4-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7 // CHECK4-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8
// CHECK4-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to double** // CHECK4-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to %struct.TT**
// CHECK4-NEXT: store double* [[VLA1]], double** [[TMP130]], align 4 // CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP130]], align 4
// CHECK4-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7 // CHECK4-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK4-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to double** // CHECK4-NEXT: store i8* null, i8** [[TMP131]], align 4
// CHECK4-NEXT: store double* [[VLA1]], double** [[TMP132]], align 4 // CHECK4-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK4-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK4-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to i32*
// CHECK4-NEXT: store i64 [[TMP86]], i64* [[TMP133]], align 4 // CHECK4-NEXT: store i32 [[TMP80]], i32* [[TMP133]], align 4
// CHECK4-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7 // CHECK4-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK4-NEXT: store i8* null, i8** [[TMP134]], align 4 // CHECK4-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i32*
// CHECK4-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8 // CHECK4-NEXT: store i32 [[TMP80]], i32* [[TMP135]], align 4
// CHECK4-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT** // CHECK4-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP136]], align 4
// CHECK4-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8 // CHECK4-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK4-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT** // CHECK4-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 4 // CHECK4-NEXT: [[TMP139:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP137]], i8** [[TMP138]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK4-NEXT: [[TMP140:%.*]] = icmp ne i32 [[TMP139]], 0
// CHECK4-NEXT: store i64 12, i64* [[TMP139]], align 4 // CHECK4-NEXT: br i1 [[TMP140]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK4-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK4-NEXT: store i8* null, i8** [[TMP140]], align 4
// CHECK4-NEXT: [[TMP141:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK4-NEXT: [[TMP142:%.*]] = bitcast i8** [[TMP141]] to i32*
// CHECK4-NEXT: store i32 [[TMP80]], i32* [[TMP142]], align 4
// CHECK4-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK4-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i32*
// CHECK4-NEXT: store i32 [[TMP80]], i32* [[TMP144]], align 4
// CHECK4-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK4-NEXT: store i64 4, i64* [[TMP145]], align 4
// CHECK4-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK4-NEXT: store i8* null, i8** [[TMP146]], align 4
// CHECK4-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK4-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK4-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK4-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP147]], i8** [[TMP148]], i64* [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK4-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK4: omp_offload.failed17: // CHECK4: omp_offload.failed17:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT18]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK4: omp_offload.cont18: // CHECK4: omp_offload.cont18:
// CHECK4-NEXT: br label [[OMP_IF_END20:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END20:%.*]]
// CHECK4: omp_if.else19: // CHECK4: omp_if.else19:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_IF_END20]] // CHECK4-NEXT: br label [[OMP_IF_END20]]
// CHECK4: omp_if.end20: // CHECK4: omp_if.end20:
// CHECK4-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4 // CHECK4-NEXT: [[TMP141:%.*]] = load i32, i32* [[A]], align 4
// CHECK4-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: [[TMP142:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP153]]) // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP142]])
// CHECK4-NEXT: ret i32 [[TMP152]] // CHECK4-NEXT: ret i32 [[TMP141]]
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK4-SAME: () #[[ATTR2:[0-9]+]] { // CHECK4-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
▲ Show 20 Lines • Show All 734 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK4-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK4-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK4-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 13 Lines
// CHECK4-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK4-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK4-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK4-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK4-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK4-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK4-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK4-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK4-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK4-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK4-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK4-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK4-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK4-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK4-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK4-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK4-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK4-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK4-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK4-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK4-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK4-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK4-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK4-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK4-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK4-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK4-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK4-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK4-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK4-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK4-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK4-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK4-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK4-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK4-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK4-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK4-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK4-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK4-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK4-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK4-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK4-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK4-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK4-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK4-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4 // CHECK4-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK4-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK4-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK4-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK4-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK4-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK4-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK4-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_IF_END]] // CHECK4-NEXT: br label [[OMP_IF_END]]
// CHECK4: omp_if.end: // CHECK4: omp_if.end:
// CHECK4-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK4-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK4-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK4-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK4-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK4-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK4-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK4-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK4-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK4-NEXT: ret i32 [[ADD3]] // CHECK4-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK4-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK4-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK4-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK4-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK4-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK4-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK4-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK4-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK4-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
Show All 20 Lines
// CHECK4-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK4-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK4-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK4-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 99 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK4-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK4-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK4-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK4-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK4-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK4-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK4-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK4-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
Show All 26 Lines
// CHECK4-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK4-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK4-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 3,278 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8
// CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK17-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK17-NEXT: store i32 0, i32* [[A]], align 4 // CHECK17-NEXT: store i32 0, i32* [[A]], align 4
// CHECK17-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK17-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK17-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK17-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK17-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK17-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK17-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 153 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[TMP87:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK17-NEXT: [[TMP87:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK17-NEXT: [[TMP88:%.*]] = mul nuw i64 [[TMP87]], 8 // CHECK17-NEXT: [[TMP88:%.*]] = mul nuw i64 [[TMP87]], 8
// CHECK17-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK17-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK17-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK17-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64*
// CHECK17-NEXT: store i64 [[TMP82]], i64* [[TMP90]], align 8 // CHECK17-NEXT: store i64 [[TMP82]], i64* [[TMP90]], align 8
// CHECK17-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK17-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK17-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i64* // CHECK17-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i64*
// CHECK17-NEXT: store i64 [[TMP82]], i64* [[TMP92]], align 8 // CHECK17-NEXT: store i64 [[TMP82]], i64* [[TMP92]], align 8
// CHECK17-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK17-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK17-NEXT: store i64 4, i64* [[TMP93]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP93]], align 8
// CHECK17-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK17-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK17-NEXT: store i8* null, i8** [[TMP94]], align 8 // CHECK17-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [10 x float]**
// CHECK17-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP95]], align 8
// CHECK17-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to [10 x float]** // CHECK17-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP96]], align 8 // CHECK17-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to [10 x float]**
// CHECK17-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP97]], align 8
// CHECK17-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to [10 x float]** // CHECK17-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP98]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP98]], align 8
// CHECK17-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK17-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK17-NEXT: store i64 40, i64* [[TMP99]], align 8 // CHECK17-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i64*
// CHECK17-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP100]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP100]], align 8 // CHECK17-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK17-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK17-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i64* // CHECK17-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i64*
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP102]], align 8 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP102]], align 8
// CHECK17-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK17-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK17-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to i64* // CHECK17-NEXT: store i8* null, i8** [[TMP103]], align 8
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP104]], align 8 // CHECK17-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK17-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK17-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to float**
// CHECK17-NEXT: store i64 8, i64* [[TMP105]], align 8 // CHECK17-NEXT: store float* [[VLA]], float** [[TMP105]], align 8
// CHECK17-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK17-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK17-NEXT: store i8* null, i8** [[TMP106]], align 8 // CHECK17-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to float**
// CHECK17-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3 // CHECK17-NEXT: store float* [[VLA]], float** [[TMP107]], align 8
// CHECK17-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to float** // CHECK17-NEXT: store i64 [[TMP86]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK17-NEXT: store float* [[VLA]], float** [[TMP108]], align 8 // CHECK17-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK17-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3 // CHECK17-NEXT: store i8* null, i8** [[TMP108]], align 8
// CHECK17-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to float** // CHECK17-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4
// CHECK17-NEXT: store float* [[VLA]], float** [[TMP110]], align 8 // CHECK17-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to [5 x [10 x double]]**
// CHECK17-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP110]], align 8
// CHECK17-NEXT: store i64 [[TMP86]], i64* [[TMP111]], align 8 // CHECK17-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4
// CHECK17-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3 // CHECK17-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to [5 x [10 x double]]**
// CHECK17-NEXT: store i8* null, i8** [[TMP112]], align 8 // CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP112]], align 8
// CHECK17-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4 // CHECK17-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4
// CHECK17-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to [5 x [10 x double]]** // CHECK17-NEXT: store i8* null, i8** [[TMP113]], align 8
// CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP114]], align 8 // CHECK17-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5
// CHECK17-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4 // CHECK17-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to i64*
// CHECK17-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]** // CHECK17-NEXT: store i64 5, i64* [[TMP115]], align 8
// CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 8 // CHECK17-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5
// CHECK17-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK17-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to i64*
// CHECK17-NEXT: store i64 400, i64* [[TMP117]], align 8 // CHECK17-NEXT: store i64 5, i64* [[TMP117]], align 8
// CHECK17-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4 // CHECK17-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5
// CHECK17-NEXT: store i8* null, i8** [[TMP118]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP118]], align 8
// CHECK17-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5 // CHECK17-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6
// CHECK17-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i64* // CHECK17-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i64*
// CHECK17-NEXT: store i64 5, i64* [[TMP120]], align 8 // CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP120]], align 8
// CHECK17-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5 // CHECK17-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6
// CHECK17-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to i64* // CHECK17-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to i64*
// CHECK17-NEXT: store i64 5, i64* [[TMP122]], align 8 // CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP122]], align 8
// CHECK17-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK17-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6
// CHECK17-NEXT: store i64 8, i64* [[TMP123]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP123]], align 8
// CHECK17-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5 // CHECK17-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7
// CHECK17-NEXT: store i8* null, i8** [[TMP124]], align 8 // CHECK17-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to double**
// CHECK17-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6 // CHECK17-NEXT: store double* [[VLA1]], double** [[TMP125]], align 8
// CHECK17-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64* // CHECK17-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7
// CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP126]], align 8 // CHECK17-NEXT: [[TMP127:%.*]] = bitcast i8** [[TMP126]] to double**
// CHECK17-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6 // CHECK17-NEXT: store double* [[VLA1]], double** [[TMP127]], align 8
// CHECK17-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64* // CHECK17-NEXT: store i64 [[TMP88]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP128]], align 8 // CHECK17-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7
// CHECK17-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK17-NEXT: store i8* null, i8** [[TMP128]], align 8
// CHECK17-NEXT: store i64 8, i64* [[TMP129]], align 8 // CHECK17-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8
// CHECK17-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6 // CHECK17-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to %struct.TT**
// CHECK17-NEXT: store i8* null, i8** [[TMP130]], align 8 // CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP130]], align 8
// CHECK17-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7 // CHECK17-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8
// CHECK17-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to double** // CHECK17-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to %struct.TT**
// CHECK17-NEXT: store double* [[VLA1]], double** [[TMP132]], align 8 // CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP132]], align 8
// CHECK17-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7 // CHECK17-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK17-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double** // CHECK17-NEXT: store i8* null, i8** [[TMP133]], align 8
// CHECK17-NEXT: store double* [[VLA1]], double** [[TMP134]], align 8 // CHECK17-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK17-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK17-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i64*
// CHECK17-NEXT: store i64 [[TMP88]], i64* [[TMP135]], align 8 // CHECK17-NEXT: store i64 [[TMP84]], i64* [[TMP135]], align 8
// CHECK17-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7 // CHECK17-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK17-NEXT: store i8* null, i8** [[TMP136]], align 8 // CHECK17-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i64*
// CHECK17-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8 // CHECK17-NEXT: store i64 [[TMP84]], i64* [[TMP137]], align 8
// CHECK17-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT** // CHECK17-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP138]], align 8
// CHECK17-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8 // CHECK17-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK17-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to %struct.TT** // CHECK17-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP140]], align 8 // CHECK17-NEXT: [[TMP141:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP139]], i8** [[TMP140]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: [[TMP141:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK17-NEXT: [[TMP142:%.*]] = icmp ne i32 [[TMP141]], 0
// CHECK17-NEXT: store i64 16, i64* [[TMP141]], align 8 // CHECK17-NEXT: br i1 [[TMP142]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK17-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK17-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK17-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK17-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i64*
// CHECK17-NEXT: store i64 [[TMP84]], i64* [[TMP144]], align 8
// CHECK17-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK17-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i64*
// CHECK17-NEXT: store i64 [[TMP84]], i64* [[TMP146]], align 8
// CHECK17-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK17-NEXT: store i64 4, i64* [[TMP147]], align 8
// CHECK17-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK17-NEXT: store i8* null, i8** [[TMP148]], align 8
// CHECK17-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK17-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK17-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK17-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0
// CHECK17-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK17: omp_offload.failed23: // CHECK17: omp_offload.failed23:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT24]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT24]]
// CHECK17: omp_offload.cont24: // CHECK17: omp_offload.cont24:
// CHECK17-NEXT: br label [[OMP_IF_END26:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END26:%.*]]
// CHECK17: omp_if.else25: // CHECK17: omp_if.else25:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_IF_END26]] // CHECK17-NEXT: br label [[OMP_IF_END26]]
// CHECK17: omp_if.end26: // CHECK17: omp_if.end26:
// CHECK17-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK17-NEXT: [[TMP143:%.*]] = load i32, i32* [[A]], align 4
// CHECK17-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK17-NEXT: [[TMP144:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP144]])
// CHECK17-NEXT: ret i32 [[TMP154]] // CHECK17-NEXT: ret i32 [[TMP143]]
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK17-SAME: () #[[ATTR2:[0-9]+]] { // CHECK17-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
▲ Show 20 Lines • Show All 756 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK17-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK17-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK17-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK17-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK17-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK17-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 14 Lines
// CHECK17-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK17-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK17-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK17-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK17-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK17-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK17-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK17-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK17-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK17-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK17-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK17-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK17-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK17-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK17-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK17-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK17-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK17-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK17-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK17-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK17-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK17-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK17-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK17-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK17-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK17-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK17-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK17-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK17-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK17-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK17-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK17-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK17-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK17-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK17-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK17-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK17-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK17-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK17-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK17-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK17-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK17-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8 // CHECK17-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK17-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK17-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK17-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK17-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK17-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK17-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK17-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: br label [[OMP_IF_END:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK17: omp_if.else: // CHECK17: omp_if.else:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_IF_END]] // CHECK17-NEXT: br label [[OMP_IF_END]]
// CHECK17: omp_if.end: // CHECK17: omp_if.end:
// CHECK17-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK17-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK17-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK17-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK17-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK17-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK17-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK17-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK17-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK17-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK17-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK17-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK17-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK17-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK17-NEXT: ret i32 [[ADD4]] // CHECK17-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK17-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK17-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK17-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK17-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK17-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK17-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK17-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK17-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK17-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK17-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: br label [[OMP_IF_END:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK17: omp_if.else: // CHECK17: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK17-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK17-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK17-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK17-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK17-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK17-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: br label [[OMP_IF_END:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK17: omp_if.else: // CHECK17: omp_if.else:
Show All 22 Lines
// CHECK17-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK17-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK17-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK17-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK17-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK17-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK17-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK17-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK17-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 102 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK17-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK17-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK17-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK17-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK17-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK17-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK17-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK17-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK17-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK17-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK17-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK17-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK17-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK17-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK17-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK17-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
Show All 29 Lines
// CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK17-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK17-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK17-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK17-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK17-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK17-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK17-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK17-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK17-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK17-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK17-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 111 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8
// CHECK18-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK18-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK18-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK18-NEXT: store i32 0, i32* [[A]], align 4 // CHECK18-NEXT: store i32 0, i32* [[A]], align 4
// CHECK18-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK18-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK18-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK18-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK18-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK18-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK18-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 153 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[TMP87:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK18-NEXT: [[TMP87:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK18-NEXT: [[TMP88:%.*]] = mul nuw i64 [[TMP87]], 8 // CHECK18-NEXT: [[TMP88:%.*]] = mul nuw i64 [[TMP87]], 8
// CHECK18-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK18-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK18-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK18-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64*
// CHECK18-NEXT: store i64 [[TMP82]], i64* [[TMP90]], align 8 // CHECK18-NEXT: store i64 [[TMP82]], i64* [[TMP90]], align 8
// CHECK18-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK18-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK18-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i64* // CHECK18-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i64*
// CHECK18-NEXT: store i64 [[TMP82]], i64* [[TMP92]], align 8 // CHECK18-NEXT: store i64 [[TMP82]], i64* [[TMP92]], align 8
// CHECK18-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK18-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK18-NEXT: store i64 4, i64* [[TMP93]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP93]], align 8
// CHECK18-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK18-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK18-NEXT: store i8* null, i8** [[TMP94]], align 8 // CHECK18-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [10 x float]**
// CHECK18-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP95]], align 8
// CHECK18-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to [10 x float]** // CHECK18-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP96]], align 8 // CHECK18-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to [10 x float]**
// CHECK18-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP97]], align 8
// CHECK18-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to [10 x float]** // CHECK18-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP98]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP98]], align 8
// CHECK18-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK18-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK18-NEXT: store i64 40, i64* [[TMP99]], align 8 // CHECK18-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i64*
// CHECK18-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP100]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP100]], align 8 // CHECK18-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK18-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK18-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i64* // CHECK18-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i64*
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP102]], align 8 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP102]], align 8
// CHECK18-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK18-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK18-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to i64* // CHECK18-NEXT: store i8* null, i8** [[TMP103]], align 8
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP104]], align 8 // CHECK18-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK18-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK18-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to float**
// CHECK18-NEXT: store i64 8, i64* [[TMP105]], align 8 // CHECK18-NEXT: store float* [[VLA]], float** [[TMP105]], align 8
// CHECK18-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK18-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK18-NEXT: store i8* null, i8** [[TMP106]], align 8 // CHECK18-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to float**
// CHECK18-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3 // CHECK18-NEXT: store float* [[VLA]], float** [[TMP107]], align 8
// CHECK18-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to float** // CHECK18-NEXT: store i64 [[TMP86]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK18-NEXT: store float* [[VLA]], float** [[TMP108]], align 8 // CHECK18-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK18-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3 // CHECK18-NEXT: store i8* null, i8** [[TMP108]], align 8
// CHECK18-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to float** // CHECK18-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4
// CHECK18-NEXT: store float* [[VLA]], float** [[TMP110]], align 8 // CHECK18-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to [5 x [10 x double]]**
// CHECK18-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP110]], align 8
// CHECK18-NEXT: store i64 [[TMP86]], i64* [[TMP111]], align 8 // CHECK18-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4
// CHECK18-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3 // CHECK18-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to [5 x [10 x double]]**
// CHECK18-NEXT: store i8* null, i8** [[TMP112]], align 8 // CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP112]], align 8
// CHECK18-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4 // CHECK18-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4
// CHECK18-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to [5 x [10 x double]]** // CHECK18-NEXT: store i8* null, i8** [[TMP113]], align 8
// CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP114]], align 8 // CHECK18-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5
// CHECK18-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4 // CHECK18-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to i64*
// CHECK18-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]** // CHECK18-NEXT: store i64 5, i64* [[TMP115]], align 8
// CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 8 // CHECK18-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5
// CHECK18-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK18-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to i64*
// CHECK18-NEXT: store i64 400, i64* [[TMP117]], align 8 // CHECK18-NEXT: store i64 5, i64* [[TMP117]], align 8
// CHECK18-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4 // CHECK18-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5
// CHECK18-NEXT: store i8* null, i8** [[TMP118]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP118]], align 8
// CHECK18-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5 // CHECK18-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6
// CHECK18-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i64* // CHECK18-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i64*
// CHECK18-NEXT: store i64 5, i64* [[TMP120]], align 8 // CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP120]], align 8
// CHECK18-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5 // CHECK18-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6
// CHECK18-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to i64* // CHECK18-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to i64*
// CHECK18-NEXT: store i64 5, i64* [[TMP122]], align 8 // CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP122]], align 8
// CHECK18-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK18-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6
// CHECK18-NEXT: store i64 8, i64* [[TMP123]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP123]], align 8
// CHECK18-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5 // CHECK18-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7
// CHECK18-NEXT: store i8* null, i8** [[TMP124]], align 8 // CHECK18-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to double**
// CHECK18-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6 // CHECK18-NEXT: store double* [[VLA1]], double** [[TMP125]], align 8
// CHECK18-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64* // CHECK18-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7
// CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP126]], align 8 // CHECK18-NEXT: [[TMP127:%.*]] = bitcast i8** [[TMP126]] to double**
// CHECK18-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6 // CHECK18-NEXT: store double* [[VLA1]], double** [[TMP127]], align 8
// CHECK18-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64* // CHECK18-NEXT: store i64 [[TMP88]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP128]], align 8 // CHECK18-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7
// CHECK18-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK18-NEXT: store i8* null, i8** [[TMP128]], align 8
// CHECK18-NEXT: store i64 8, i64* [[TMP129]], align 8 // CHECK18-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8
// CHECK18-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6 // CHECK18-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to %struct.TT**
// CHECK18-NEXT: store i8* null, i8** [[TMP130]], align 8 // CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP130]], align 8
// CHECK18-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7 // CHECK18-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8
// CHECK18-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to double** // CHECK18-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to %struct.TT**
// CHECK18-NEXT: store double* [[VLA1]], double** [[TMP132]], align 8 // CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP132]], align 8
// CHECK18-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7 // CHECK18-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK18-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double** // CHECK18-NEXT: store i8* null, i8** [[TMP133]], align 8
// CHECK18-NEXT: store double* [[VLA1]], double** [[TMP134]], align 8 // CHECK18-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK18-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK18-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i64*
// CHECK18-NEXT: store i64 [[TMP88]], i64* [[TMP135]], align 8 // CHECK18-NEXT: store i64 [[TMP84]], i64* [[TMP135]], align 8
// CHECK18-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7 // CHECK18-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK18-NEXT: store i8* null, i8** [[TMP136]], align 8 // CHECK18-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i64*
// CHECK18-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8 // CHECK18-NEXT: store i64 [[TMP84]], i64* [[TMP137]], align 8
// CHECK18-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT** // CHECK18-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP138]], align 8
// CHECK18-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8 // CHECK18-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK18-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to %struct.TT** // CHECK18-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP140]], align 8 // CHECK18-NEXT: [[TMP141:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP139]], i8** [[TMP140]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: [[TMP141:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK18-NEXT: [[TMP142:%.*]] = icmp ne i32 [[TMP141]], 0
// CHECK18-NEXT: store i64 16, i64* [[TMP141]], align 8 // CHECK18-NEXT: br i1 [[TMP142]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK18-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK18-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK18-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK18-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i64*
// CHECK18-NEXT: store i64 [[TMP84]], i64* [[TMP144]], align 8
// CHECK18-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK18-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i64*
// CHECK18-NEXT: store i64 [[TMP84]], i64* [[TMP146]], align 8
// CHECK18-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK18-NEXT: store i64 4, i64* [[TMP147]], align 8
// CHECK18-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK18-NEXT: store i8* null, i8** [[TMP148]], align 8
// CHECK18-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK18-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK18-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK18-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0
// CHECK18-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK18: omp_offload.failed23: // CHECK18: omp_offload.failed23:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT24]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT24]]
// CHECK18: omp_offload.cont24: // CHECK18: omp_offload.cont24:
// CHECK18-NEXT: br label [[OMP_IF_END26:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END26:%.*]]
// CHECK18: omp_if.else25: // CHECK18: omp_if.else25:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i64 [[TMP82]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP84]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_IF_END26]] // CHECK18-NEXT: br label [[OMP_IF_END26]]
// CHECK18: omp_if.end26: // CHECK18: omp_if.end26:
// CHECK18-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK18-NEXT: [[TMP143:%.*]] = load i32, i32* [[A]], align 4
// CHECK18-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK18-NEXT: [[TMP144:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP144]])
// CHECK18-NEXT: ret i32 [[TMP154]] // CHECK18-NEXT: ret i32 [[TMP143]]
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK18-SAME: () #[[ATTR2:[0-9]+]] { // CHECK18-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
▲ Show 20 Lines • Show All 756 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK18-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK18-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK18-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK18-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK18-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK18-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK18-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK18-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK18-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK18-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 14 Lines
// CHECK18-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK18-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK18-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK18-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK18-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK18-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK18-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK18-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK18-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK18-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK18-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK18-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK18-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK18-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK18-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK18-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK18-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK18-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK18-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK18-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK18-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK18-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK18-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK18-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK18-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK18-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK18-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK18-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK18-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK18-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK18-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK18-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK18-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK18-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK18-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK18-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK18-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK18-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK18-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK18-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK18-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK18-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK18-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK18-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK18-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK18-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK18-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK18-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK18-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK18-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK18-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK18-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK18-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK18-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK18-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK18-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK18-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8 // CHECK18-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK18-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK18-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK18-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK18-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK18-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK18-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK18-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: br label [[OMP_IF_END:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK18: omp_if.else: // CHECK18: omp_if.else:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_IF_END]] // CHECK18-NEXT: br label [[OMP_IF_END]]
// CHECK18: omp_if.end: // CHECK18: omp_if.end:
// CHECK18-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK18-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK18-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK18-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK18-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK18-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK18-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK18-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK18-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK18-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK18-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK18-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK18-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK18-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK18-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK18-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK18-NEXT: ret i32 [[ADD4]] // CHECK18-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK18-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK18-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK18-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK18-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK18-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK18-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK18-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK18-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK18-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK18-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: br label [[OMP_IF_END:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK18: omp_if.else: // CHECK18: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK18-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK18-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK18-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK18-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK18-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK18-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: br label [[OMP_IF_END:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK18: omp_if.else: // CHECK18: omp_if.else:
Show All 22 Lines
// CHECK18-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK18-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK18-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK18-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK18-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK18-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK18-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK18-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK18-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK18-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK18-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK18-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK18-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK18-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 102 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK18-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK18-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK18-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK18-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK18-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK18-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK18-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK18-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK18-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK18-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK18-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK18-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK18-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK18-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK18-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK18-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
Show All 29 Lines
// CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK18-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK18-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK18-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK18-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK18-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK18-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK18-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK18-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK18-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK18-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK18-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK18-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4
// CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK19-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK19-NEXT: store i32 0, i32* [[A]], align 4 // CHECK19-NEXT: store i32 0, i32* [[A]], align 4
// CHECK19-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK19-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK19-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK19-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK19-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK19-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 144 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[TMP85:%.*]] = mul nuw i32 [[TMP84]], 8 // CHECK19-NEXT: [[TMP85:%.*]] = mul nuw i32 [[TMP84]], 8
// CHECK19-NEXT: [[TMP86:%.*]] = sext i32 [[TMP85]] to i64 // CHECK19-NEXT: [[TMP86:%.*]] = sext i32 [[TMP85]] to i64
// CHECK19-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0 // CHECK19-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK19-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32* // CHECK19-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32*
// CHECK19-NEXT: store i32 [[TMP78]], i32* [[TMP88]], align 4 // CHECK19-NEXT: store i32 [[TMP78]], i32* [[TMP88]], align 4
// CHECK19-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0 // CHECK19-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK19-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK19-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32*
// CHECK19-NEXT: store i32 [[TMP78]], i32* [[TMP90]], align 4 // CHECK19-NEXT: store i32 [[TMP78]], i32* [[TMP90]], align 4
// CHECK19-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK19-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0
// CHECK19-NEXT: store i64 4, i64* [[TMP91]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP91]], align 4
// CHECK19-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0 // CHECK19-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1
// CHECK19-NEXT: store i8* null, i8** [[TMP92]], align 4 // CHECK19-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [10 x float]**
// CHECK19-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1 // CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP93]], align 4
// CHECK19-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to [10 x float]** // CHECK19-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1
// CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP94]], align 4 // CHECK19-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [10 x float]**
// CHECK19-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1 // CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP95]], align 4
// CHECK19-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to [10 x float]** // CHECK19-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1
// CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP96]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP96]], align 4
// CHECK19-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK19-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK19-NEXT: store i64 40, i64* [[TMP97]], align 4 // CHECK19-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32*
// CHECK19-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP98]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP98]], align 4 // CHECK19-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2
// CHECK19-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK19-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i32* // CHECK19-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i32*
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP100]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP100]], align 4
// CHECK19-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2 // CHECK19-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2
// CHECK19-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i32* // CHECK19-NEXT: store i8* null, i8** [[TMP101]], align 4
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP102]], align 4 // CHECK19-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3
// CHECK19-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK19-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to float**
// CHECK19-NEXT: store i64 4, i64* [[TMP103]], align 4 // CHECK19-NEXT: store float* [[VLA]], float** [[TMP103]], align 4
// CHECK19-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2 // CHECK19-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3
// CHECK19-NEXT: store i8* null, i8** [[TMP104]], align 4 // CHECK19-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to float**
// CHECK19-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3 // CHECK19-NEXT: store float* [[VLA]], float** [[TMP105]], align 4
// CHECK19-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to float** // CHECK19-NEXT: store i64 [[TMP83]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK19-NEXT: store float* [[VLA]], float** [[TMP106]], align 4 // CHECK19-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3
// CHECK19-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3 // CHECK19-NEXT: store i8* null, i8** [[TMP106]], align 4
// CHECK19-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to float** // CHECK19-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4
// CHECK19-NEXT: store float* [[VLA]], float** [[TMP108]], align 4 // CHECK19-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to [5 x [10 x double]]**
// CHECK19-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP108]], align 4
// CHECK19-NEXT: store i64 [[TMP83]], i64* [[TMP109]], align 4 // CHECK19-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4
// CHECK19-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3 // CHECK19-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to [5 x [10 x double]]**
// CHECK19-NEXT: store i8* null, i8** [[TMP110]], align 4 // CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP110]], align 4
// CHECK19-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4 // CHECK19-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4
// CHECK19-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to [5 x [10 x double]]** // CHECK19-NEXT: store i8* null, i8** [[TMP111]], align 4
// CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP112]], align 4 // CHECK19-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5
// CHECK19-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4 // CHECK19-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i32*
// CHECK19-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to [5 x [10 x double]]** // CHECK19-NEXT: store i32 5, i32* [[TMP113]], align 4
// CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP114]], align 4 // CHECK19-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5
// CHECK19-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK19-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to i32*
// CHECK19-NEXT: store i64 400, i64* [[TMP115]], align 4 // CHECK19-NEXT: store i32 5, i32* [[TMP115]], align 4
// CHECK19-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4 // CHECK19-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5
// CHECK19-NEXT: store i8* null, i8** [[TMP116]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP116]], align 4
// CHECK19-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5 // CHECK19-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6
// CHECK19-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to i32* // CHECK19-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to i32*
// CHECK19-NEXT: store i32 5, i32* [[TMP118]], align 4 // CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP118]], align 4
// CHECK19-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5 // CHECK19-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6
// CHECK19-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i32* // CHECK19-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i32*
// CHECK19-NEXT: store i32 5, i32* [[TMP120]], align 4 // CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP120]], align 4
// CHECK19-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK19-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6
// CHECK19-NEXT: store i64 4, i64* [[TMP121]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP121]], align 4
// CHECK19-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5 // CHECK19-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7
// CHECK19-NEXT: store i8* null, i8** [[TMP122]], align 4 // CHECK19-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to double**
// CHECK19-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6 // CHECK19-NEXT: store double* [[VLA1]], double** [[TMP123]], align 4
// CHECK19-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i32* // CHECK19-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7
// CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP124]], align 4 // CHECK19-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to double**
// CHECK19-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6 // CHECK19-NEXT: store double* [[VLA1]], double** [[TMP125]], align 4
// CHECK19-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32* // CHECK19-NEXT: store i64 [[TMP86]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP126]], align 4 // CHECK19-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7
// CHECK19-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK19-NEXT: store i8* null, i8** [[TMP126]], align 4
// CHECK19-NEXT: store i64 4, i64* [[TMP127]], align 4 // CHECK19-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8
// CHECK19-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6 // CHECK19-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to %struct.TT**
// CHECK19-NEXT: store i8* null, i8** [[TMP128]], align 4 // CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP128]], align 4
// CHECK19-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7 // CHECK19-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8
// CHECK19-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to double** // CHECK19-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to %struct.TT**
// CHECK19-NEXT: store double* [[VLA1]], double** [[TMP130]], align 4 // CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP130]], align 4
// CHECK19-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7 // CHECK19-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK19-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to double** // CHECK19-NEXT: store i8* null, i8** [[TMP131]], align 4
// CHECK19-NEXT: store double* [[VLA1]], double** [[TMP132]], align 4 // CHECK19-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK19-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK19-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to i32*
// CHECK19-NEXT: store i64 [[TMP86]], i64* [[TMP133]], align 4 // CHECK19-NEXT: store i32 [[TMP80]], i32* [[TMP133]], align 4
// CHECK19-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7 // CHECK19-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK19-NEXT: store i8* null, i8** [[TMP134]], align 4 // CHECK19-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i32*
// CHECK19-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8 // CHECK19-NEXT: store i32 [[TMP80]], i32* [[TMP135]], align 4
// CHECK19-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT** // CHECK19-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP136]], align 4
// CHECK19-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8 // CHECK19-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK19-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT** // CHECK19-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 4 // CHECK19-NEXT: [[TMP139:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP137]], i8** [[TMP138]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK19-NEXT: [[TMP140:%.*]] = icmp ne i32 [[TMP139]], 0
// CHECK19-NEXT: store i64 12, i64* [[TMP139]], align 4 // CHECK19-NEXT: br i1 [[TMP140]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK19-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK19-NEXT: store i8* null, i8** [[TMP140]], align 4
// CHECK19-NEXT: [[TMP141:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK19-NEXT: [[TMP142:%.*]] = bitcast i8** [[TMP141]] to i32*
// CHECK19-NEXT: store i32 [[TMP80]], i32* [[TMP142]], align 4
// CHECK19-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK19-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i32*
// CHECK19-NEXT: store i32 [[TMP80]], i32* [[TMP144]], align 4
// CHECK19-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK19-NEXT: store i64 4, i64* [[TMP145]], align 4
// CHECK19-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK19-NEXT: store i8* null, i8** [[TMP146]], align 4
// CHECK19-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK19-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK19-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK19-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP147]], i8** [[TMP148]], i64* [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK19-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK19: omp_offload.failed17: // CHECK19: omp_offload.failed17:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT18]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK19: omp_offload.cont18: // CHECK19: omp_offload.cont18:
// CHECK19-NEXT: br label [[OMP_IF_END20:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END20:%.*]]
// CHECK19: omp_if.else19: // CHECK19: omp_if.else19:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_IF_END20]] // CHECK19-NEXT: br label [[OMP_IF_END20]]
// CHECK19: omp_if.end20: // CHECK19: omp_if.end20:
// CHECK19-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4 // CHECK19-NEXT: [[TMP141:%.*]] = load i32, i32* [[A]], align 4
// CHECK19-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: [[TMP142:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP153]]) // CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP142]])
// CHECK19-NEXT: ret i32 [[TMP152]] // CHECK19-NEXT: ret i32 [[TMP141]]
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK19-SAME: () #[[ATTR2:[0-9]+]] { // CHECK19-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
▲ Show 20 Lines • Show All 734 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK19-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK19-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK19-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK19-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK19-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK19-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 13 Lines
// CHECK19-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK19-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK19-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK19-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK19-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK19-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK19-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK19-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK19-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK19-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK19-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK19-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK19-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK19-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK19-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK19-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK19-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK19-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK19-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK19-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK19-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK19-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK19-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK19-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK19-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK19-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK19-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK19-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK19-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK19-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK19-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK19-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK19-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK19-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK19-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK19-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK19-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK19-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK19-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK19-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK19-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK19-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK19-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK19-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4 // CHECK19-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK19-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK19-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK19-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK19-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK19-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK19-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK19-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: br label [[OMP_IF_END:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK19: omp_if.else: // CHECK19: omp_if.else:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_IF_END]] // CHECK19-NEXT: br label [[OMP_IF_END]]
// CHECK19: omp_if.end: // CHECK19: omp_if.end:
// CHECK19-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK19-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK19-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK19-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK19-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK19-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK19-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK19-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK19-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK19-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK19-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK19-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK19-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK19-NEXT: ret i32 [[ADD3]] // CHECK19-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK19-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK19-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK19-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK19-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK19-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK19-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK19-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK19-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK19-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK19-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: br label [[OMP_IF_END:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK19: omp_if.else: // CHECK19: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK19-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK19-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK19-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK19-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK19-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK19-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK19-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: br label [[OMP_IF_END:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK19: omp_if.else: // CHECK19: omp_if.else:
Show All 20 Lines
// CHECK19-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK19-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK19-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK19-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK19-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 99 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK19-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK19-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK19-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK19-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK19-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK19-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK19-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK19-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK19-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK19-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK19-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK19-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
Show All 26 Lines
// CHECK19-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK19-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK19-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK19-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK19-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK19-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK19-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 109 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4
// CHECK20-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK20-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK20-NEXT: store i32 0, i32* [[A]], align 4 // CHECK20-NEXT: store i32 0, i32* [[A]], align 4
// CHECK20-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK20-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK20-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK20-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK20-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK20-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 144 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[TMP85:%.*]] = mul nuw i32 [[TMP84]], 8 // CHECK20-NEXT: [[TMP85:%.*]] = mul nuw i32 [[TMP84]], 8
// CHECK20-NEXT: [[TMP86:%.*]] = sext i32 [[TMP85]] to i64 // CHECK20-NEXT: [[TMP86:%.*]] = sext i32 [[TMP85]] to i64
// CHECK20-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0 // CHECK20-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK20-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32* // CHECK20-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i32*
// CHECK20-NEXT: store i32 [[TMP78]], i32* [[TMP88]], align 4 // CHECK20-NEXT: store i32 [[TMP78]], i32* [[TMP88]], align 4
// CHECK20-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0 // CHECK20-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK20-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK20-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32*
// CHECK20-NEXT: store i32 [[TMP78]], i32* [[TMP90]], align 4 // CHECK20-NEXT: store i32 [[TMP78]], i32* [[TMP90]], align 4
// CHECK20-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK20-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0
// CHECK20-NEXT: store i64 4, i64* [[TMP91]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP91]], align 4
// CHECK20-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0 // CHECK20-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1
// CHECK20-NEXT: store i8* null, i8** [[TMP92]], align 4 // CHECK20-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [10 x float]**
// CHECK20-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1 // CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP93]], align 4
// CHECK20-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to [10 x float]** // CHECK20-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1
// CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP94]], align 4 // CHECK20-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [10 x float]**
// CHECK20-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1 // CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP95]], align 4
// CHECK20-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to [10 x float]** // CHECK20-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1
// CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP96]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP96]], align 4
// CHECK20-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK20-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK20-NEXT: store i64 40, i64* [[TMP97]], align 4 // CHECK20-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32*
// CHECK20-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP98]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP98]], align 4 // CHECK20-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2
// CHECK20-NEXT: [[TMP99:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK20-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i32* // CHECK20-NEXT: [[TMP100:%.*]] = bitcast i8** [[TMP99]] to i32*
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP100]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP100]], align 4
// CHECK20-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2 // CHECK20-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2
// CHECK20-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to i32* // CHECK20-NEXT: store i8* null, i8** [[TMP101]], align 4
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP102]], align 4 // CHECK20-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3
// CHECK20-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK20-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to float**
// CHECK20-NEXT: store i64 4, i64* [[TMP103]], align 4 // CHECK20-NEXT: store float* [[VLA]], float** [[TMP103]], align 4
// CHECK20-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2 // CHECK20-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3
// CHECK20-NEXT: store i8* null, i8** [[TMP104]], align 4 // CHECK20-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to float**
// CHECK20-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3 // CHECK20-NEXT: store float* [[VLA]], float** [[TMP105]], align 4
// CHECK20-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to float** // CHECK20-NEXT: store i64 [[TMP83]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK20-NEXT: store float* [[VLA]], float** [[TMP106]], align 4 // CHECK20-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3
// CHECK20-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3 // CHECK20-NEXT: store i8* null, i8** [[TMP106]], align 4
// CHECK20-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to float** // CHECK20-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4
// CHECK20-NEXT: store float* [[VLA]], float** [[TMP108]], align 4 // CHECK20-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to [5 x [10 x double]]**
// CHECK20-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP108]], align 4
// CHECK20-NEXT: store i64 [[TMP83]], i64* [[TMP109]], align 4 // CHECK20-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4
// CHECK20-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3 // CHECK20-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to [5 x [10 x double]]**
// CHECK20-NEXT: store i8* null, i8** [[TMP110]], align 4 // CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP110]], align 4
// CHECK20-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4 // CHECK20-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4
// CHECK20-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to [5 x [10 x double]]** // CHECK20-NEXT: store i8* null, i8** [[TMP111]], align 4
// CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP112]], align 4 // CHECK20-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5
// CHECK20-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4 // CHECK20-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i32*
// CHECK20-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to [5 x [10 x double]]** // CHECK20-NEXT: store i32 5, i32* [[TMP113]], align 4
// CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP114]], align 4 // CHECK20-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5
// CHECK20-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK20-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to i32*
// CHECK20-NEXT: store i64 400, i64* [[TMP115]], align 4 // CHECK20-NEXT: store i32 5, i32* [[TMP115]], align 4
// CHECK20-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4 // CHECK20-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5
// CHECK20-NEXT: store i8* null, i8** [[TMP116]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP116]], align 4
// CHECK20-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5 // CHECK20-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6
// CHECK20-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to i32* // CHECK20-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to i32*
// CHECK20-NEXT: store i32 5, i32* [[TMP118]], align 4 // CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP118]], align 4
// CHECK20-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5 // CHECK20-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6
// CHECK20-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i32* // CHECK20-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to i32*
// CHECK20-NEXT: store i32 5, i32* [[TMP120]], align 4 // CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP120]], align 4
// CHECK20-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK20-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6
// CHECK20-NEXT: store i64 4, i64* [[TMP121]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP121]], align 4
// CHECK20-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5 // CHECK20-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7
// CHECK20-NEXT: store i8* null, i8** [[TMP122]], align 4 // CHECK20-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to double**
// CHECK20-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6 // CHECK20-NEXT: store double* [[VLA1]], double** [[TMP123]], align 4
// CHECK20-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i32* // CHECK20-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7
// CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP124]], align 4 // CHECK20-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to double**
// CHECK20-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6 // CHECK20-NEXT: store double* [[VLA1]], double** [[TMP125]], align 4
// CHECK20-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32* // CHECK20-NEXT: store i64 [[TMP86]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP126]], align 4 // CHECK20-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7
// CHECK20-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK20-NEXT: store i8* null, i8** [[TMP126]], align 4
// CHECK20-NEXT: store i64 4, i64* [[TMP127]], align 4 // CHECK20-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8
// CHECK20-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6 // CHECK20-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to %struct.TT**
// CHECK20-NEXT: store i8* null, i8** [[TMP128]], align 4 // CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP128]], align 4
// CHECK20-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7 // CHECK20-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8
// CHECK20-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to double** // CHECK20-NEXT: [[TMP130:%.*]] = bitcast i8** [[TMP129]] to %struct.TT**
// CHECK20-NEXT: store double* [[VLA1]], double** [[TMP130]], align 4 // CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP130]], align 4
// CHECK20-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7 // CHECK20-NEXT: [[TMP131:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK20-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to double** // CHECK20-NEXT: store i8* null, i8** [[TMP131]], align 4
// CHECK20-NEXT: store double* [[VLA1]], double** [[TMP132]], align 4 // CHECK20-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK20-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK20-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to i32*
// CHECK20-NEXT: store i64 [[TMP86]], i64* [[TMP133]], align 4 // CHECK20-NEXT: store i32 [[TMP80]], i32* [[TMP133]], align 4
// CHECK20-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7 // CHECK20-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK20-NEXT: store i8* null, i8** [[TMP134]], align 4 // CHECK20-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i32*
// CHECK20-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8 // CHECK20-NEXT: store i32 [[TMP80]], i32* [[TMP135]], align 4
// CHECK20-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT** // CHECK20-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP136]], align 4
// CHECK20-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8 // CHECK20-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK20-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT** // CHECK20-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 4 // CHECK20-NEXT: [[TMP139:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP137]], i8** [[TMP138]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK20-NEXT: [[TMP140:%.*]] = icmp ne i32 [[TMP139]], 0
// CHECK20-NEXT: store i64 12, i64* [[TMP139]], align 4 // CHECK20-NEXT: br i1 [[TMP140]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK20-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK20-NEXT: store i8* null, i8** [[TMP140]], align 4
// CHECK20-NEXT: [[TMP141:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK20-NEXT: [[TMP142:%.*]] = bitcast i8** [[TMP141]] to i32*
// CHECK20-NEXT: store i32 [[TMP80]], i32* [[TMP142]], align 4
// CHECK20-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK20-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i32*
// CHECK20-NEXT: store i32 [[TMP80]], i32* [[TMP144]], align 4
// CHECK20-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK20-NEXT: store i64 4, i64* [[TMP145]], align 4
// CHECK20-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK20-NEXT: store i8* null, i8** [[TMP146]], align 4
// CHECK20-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK20-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK20-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK20-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170.region_id, i32 10, i8** [[TMP147]], i8** [[TMP148]], i64* [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK20-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK20: omp_offload.failed17: // CHECK20: omp_offload.failed17:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT18]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK20: omp_offload.cont18: // CHECK20: omp_offload.cont18:
// CHECK20-NEXT: br label [[OMP_IF_END20:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END20:%.*]]
// CHECK20: omp_if.else19: // CHECK20: omp_if.else19:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l170(i32 [[TMP78]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP80]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_IF_END20]] // CHECK20-NEXT: br label [[OMP_IF_END20]]
// CHECK20: omp_if.end20: // CHECK20: omp_if.end20:
// CHECK20-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4 // CHECK20-NEXT: [[TMP141:%.*]] = load i32, i32* [[A]], align 4
// CHECK20-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: [[TMP142:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP153]]) // CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP142]])
// CHECK20-NEXT: ret i32 [[TMP152]] // CHECK20-NEXT: ret i32 [[TMP141]]
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK20-SAME: () #[[ATTR2:[0-9]+]] { // CHECK20-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
▲ Show 20 Lines • Show All 734 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK20-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK20-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK20-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK20-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK20-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK20-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK20-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK20-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK20-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 13 Lines
// CHECK20-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK20-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK20-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK20-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK20-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK20-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK20-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK20-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK20-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK20-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK20-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK20-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK20-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK20-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK20-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK20-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK20-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK20-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK20-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK20-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK20-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK20-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK20-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK20-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK20-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK20-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK20-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK20-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK20-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK20-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK20-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK20-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK20-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK20-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK20-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK20-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK20-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK20-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK20-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK20-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK20-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK20-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK20-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK20-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK20-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK20-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK20-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK20-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK20-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK20-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK20-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK20-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK20-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK20-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK20-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK20-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK20-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4 // CHECK20-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK20-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK20-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK20-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK20-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK20-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK20-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK20-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: br label [[OMP_IF_END:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK20: omp_if.else: // CHECK20: omp_if.else:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l242(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_IF_END]] // CHECK20-NEXT: br label [[OMP_IF_END]]
// CHECK20: omp_if.end: // CHECK20: omp_if.end:
// CHECK20-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK20-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK20-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK20-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK20-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK20-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK20-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK20-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK20-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK20-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK20-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK20-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK20-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK20-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK20-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK20-NEXT: ret i32 [[ADD3]] // CHECK20-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK20-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK20-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK20-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK20-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK20-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK20-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK20-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK20-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK20-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK20-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK20-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK20-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l224(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: br label [[OMP_IF_END:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK20: omp_if.else: // CHECK20: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK20-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK20-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK20-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK20-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK20-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK20-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK20-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l207(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: br label [[OMP_IF_END:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK20: omp_if.else: // CHECK20: omp_if.else:
Show All 20 Lines
// CHECK20-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK20-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK20-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK20-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK20-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK20-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK20-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK20-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK20-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK20-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 99 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK20-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK20-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK20-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK20-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK20-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK20-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK20-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK20-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK20-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK20-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK20-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK20-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK20-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK20-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
Show All 26 Lines
// CHECK20-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK20-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK20-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK20-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK20-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK20-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK20-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK20-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK20-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK20-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 3,245 Lines • Show Last 20 Lines

clang/test/OpenMP/target_parallel_for_simd_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 327 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 0, i32* [[A]], align 4 // CHECK1-NEXT: store i32 0, i32* [[A]], align 4
// CHECK1-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK1-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK1-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK1-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 124 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP66:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK1-NEXT: [[TMP66:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK1-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP66]], 8 // CHECK1-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP66]], 8
// CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK1-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64* // CHECK1-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK1-NEXT: store i64 [[TMP61]], i64* [[TMP69]], align 8 // CHECK1-NEXT: store i64 [[TMP61]], i64* [[TMP69]], align 8
// CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK1-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64* // CHECK1-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64*
// CHECK1-NEXT: store i64 [[TMP61]], i64* [[TMP71]], align 8 // CHECK1-NEXT: store i64 [[TMP61]], i64* [[TMP71]], align 8
// CHECK1-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK1-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK1-NEXT: store i64 4, i64* [[TMP72]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP72]], align 8
// CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK1-NEXT: store i8* null, i8** [[TMP73]], align 8 // CHECK1-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [10 x float]**
// CHECK1-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP74]], align 8
// CHECK1-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to [10 x float]** // CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP75]], align 8 // CHECK1-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [10 x float]**
// CHECK1-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP76]], align 8
// CHECK1-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to [10 x float]** // CHECK1-NEXT: [[TMP77:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP77]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP77]], align 8
// CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK1-NEXT: store i64 40, i64* [[TMP78]], align 8 // CHECK1-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK1-NEXT: [[TMP79:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP79]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP79]], align 8 // CHECK1-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK1-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK1-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64* // CHECK1-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64*
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP81]], align 8 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP81]], align 8
// CHECK1-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK1-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK1-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i64* // CHECK1-NEXT: store i8* null, i8** [[TMP82]], align 8
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP83]], align 8 // CHECK1-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK1-NEXT: [[TMP84:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK1-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to float**
// CHECK1-NEXT: store i64 8, i64* [[TMP84]], align 8 // CHECK1-NEXT: store float* [[VLA]], float** [[TMP84]], align 8
// CHECK1-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK1-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK1-NEXT: store i8* null, i8** [[TMP85]], align 8 // CHECK1-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to float**
// CHECK1-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3 // CHECK1-NEXT: store float* [[VLA]], float** [[TMP86]], align 8
// CHECK1-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to float** // CHECK1-NEXT: store i64 [[TMP65]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK1-NEXT: store float* [[VLA]], float** [[TMP87]], align 8 // CHECK1-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3 // CHECK1-NEXT: store i8* null, i8** [[TMP87]], align 8
// CHECK1-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to float** // CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4
// CHECK1-NEXT: store float* [[VLA]], float** [[TMP89]], align 8 // CHECK1-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to [5 x [10 x double]]**
// CHECK1-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP89]], align 8
// CHECK1-NEXT: store i64 [[TMP65]], i64* [[TMP90]], align 8 // CHECK1-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4
// CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3 // CHECK1-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to [5 x [10 x double]]**
// CHECK1-NEXT: store i8* null, i8** [[TMP91]], align 8 // CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP91]], align 8
// CHECK1-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4 // CHECK1-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4
// CHECK1-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [5 x [10 x double]]** // CHECK1-NEXT: store i8* null, i8** [[TMP92]], align 8
// CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP93]], align 8 // CHECK1-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5
// CHECK1-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4 // CHECK1-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i64*
// CHECK1-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [5 x [10 x double]]** // CHECK1-NEXT: store i64 5, i64* [[TMP94]], align 8
// CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP95]], align 8 // CHECK1-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5
// CHECK1-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK1-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64*
// CHECK1-NEXT: store i64 400, i64* [[TMP96]], align 8 // CHECK1-NEXT: store i64 5, i64* [[TMP96]], align 8
// CHECK1-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4 // CHECK1-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5
// CHECK1-NEXT: store i8* null, i8** [[TMP97]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP97]], align 8
// CHECK1-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5 // CHECK1-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6
// CHECK1-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64* // CHECK1-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64*
// CHECK1-NEXT: store i64 5, i64* [[TMP99]], align 8 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP99]], align 8
// CHECK1-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5 // CHECK1-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6
// CHECK1-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64* // CHECK1-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64*
// CHECK1-NEXT: store i64 5, i64* [[TMP101]], align 8 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP101]], align 8
// CHECK1-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK1-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6
// CHECK1-NEXT: store i64 8, i64* [[TMP102]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK1-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5 // CHECK1-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7
// CHECK1-NEXT: store i8* null, i8** [[TMP103]], align 8 // CHECK1-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to double**
// CHECK1-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6 // CHECK1-NEXT: store double* [[VLA1]], double** [[TMP104]], align 8
// CHECK1-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to i64* // CHECK1-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP105]], align 8 // CHECK1-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to double**
// CHECK1-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6 // CHECK1-NEXT: store double* [[VLA1]], double** [[TMP106]], align 8
// CHECK1-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to i64* // CHECK1-NEXT: store i64 [[TMP67]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP107]], align 8 // CHECK1-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7
// CHECK1-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK1-NEXT: store i8* null, i8** [[TMP107]], align 8
// CHECK1-NEXT: store i64 8, i64* [[TMP108]], align 8 // CHECK1-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8
// CHECK1-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6 // CHECK1-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to %struct.TT**
// CHECK1-NEXT: store i8* null, i8** [[TMP109]], align 8 // CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP109]], align 8
// CHECK1-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7 // CHECK1-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8
// CHECK1-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to double** // CHECK1-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to %struct.TT**
// CHECK1-NEXT: store double* [[VLA1]], double** [[TMP111]], align 8 // CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP111]], align 8
// CHECK1-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7 // CHECK1-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK1-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to double** // CHECK1-NEXT: store i8* null, i8** [[TMP112]], align 8
// CHECK1-NEXT: store double* [[VLA1]], double** [[TMP113]], align 8 // CHECK1-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK1-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK1-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to i64*
// CHECK1-NEXT: store i64 [[TMP67]], i64* [[TMP114]], align 8 // CHECK1-NEXT: store i64 [[TMP63]], i64* [[TMP114]], align 8
// CHECK1-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7 // CHECK1-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK1-NEXT: store i8* null, i8** [[TMP115]], align 8 // CHECK1-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to i64*
// CHECK1-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8 // CHECK1-NEXT: store i64 [[TMP63]], i64* [[TMP116]], align 8
// CHECK1-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to %struct.TT** // CHECK1-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP117]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP117]], align 8
// CHECK1-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8 // CHECK1-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK1-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to %struct.TT** // CHECK1-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP119]], align 8 // CHECK1-NEXT: [[TMP120:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP118]], i8** [[TMP119]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK1-NEXT: [[TMP121:%.*]] = icmp ne i32 [[TMP120]], 0
// CHECK1-NEXT: store i64 16, i64* [[TMP120]], align 8 // CHECK1-NEXT: br i1 [[TMP121]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK1-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK1-NEXT: store i8* null, i8** [[TMP121]], align 8
// CHECK1-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK1-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i64*
// CHECK1-NEXT: store i64 [[TMP63]], i64* [[TMP123]], align 8
// CHECK1-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK1-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to i64*
// CHECK1-NEXT: store i64 [[TMP63]], i64* [[TMP125]], align 8
// CHECK1-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK1-NEXT: store i64 4, i64* [[TMP126]], align 8
// CHECK1-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK1-NEXT: store i8* null, i8** [[TMP127]], align 8
// CHECK1-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK1-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK1-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK1-NEXT: [[TMP131:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP128]], i8** [[TMP129]], i64* [[TMP130]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP132:%.*]] = icmp ne i32 [[TMP131]], 0
// CHECK1-NEXT: br i1 [[TMP132]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK1: omp_offload.failed23: // CHECK1: omp_offload.failed23:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT24]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT24]]
// CHECK1: omp_offload.cont24: // CHECK1: omp_offload.cont24:
// CHECK1-NEXT: br label [[OMP_IF_END26:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END26:%.*]]
// CHECK1: omp_if.else25: // CHECK1: omp_if.else25:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_IF_END26]] // CHECK1-NEXT: br label [[OMP_IF_END26]]
// CHECK1: omp_if.end26: // CHECK1: omp_if.end26:
// CHECK1-NEXT: [[TMP133:%.*]] = load i32, i32* [[A]], align 4 // CHECK1-NEXT: [[TMP122:%.*]] = load i32, i32* [[A]], align 4
// CHECK1-NEXT: [[TMP134:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP123:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP134]]) // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP123]])
// CHECK1-NEXT: ret i32 [[TMP133]] // CHECK1-NEXT: ret i32 [[TMP122]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96
// CHECK1-SAME: () #[[ATTR2:[0-9]+]] { // CHECK1-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
▲ Show 20 Lines • Show All 711 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK1-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK1-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK1-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK1-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 14 Lines
// CHECK1-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK1-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK1-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK1-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK1-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK1-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK1-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK1-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK1-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK1-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK1-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK1-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK1-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK1-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK1-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK1-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK1-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK1-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK1-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK1-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK1-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK1-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8 // CHECK1-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK1-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK1-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK1-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_IF_END]] // CHECK1-NEXT: br label [[OMP_IF_END]]
// CHECK1: omp_if.end: // CHECK1: omp_if.end:
// CHECK1-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK1-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK1-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK1-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK1-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK1-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK1-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK1-NEXT: ret i32 [[ADD4]] // CHECK1-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK1-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK1-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK1-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK1-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK1-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
Show All 22 Lines
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 109 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK1-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK1-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK1-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK1-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK1-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK1-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK1-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
Show All 29 Lines
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 118 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 0, i32* [[A]], align 4 // CHECK2-NEXT: store i32 0, i32* [[A]], align 4
// CHECK2-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK2-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK2-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK2-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 124 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP66:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK2-NEXT: [[TMP66:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK2-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP66]], 8 // CHECK2-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP66]], 8
// CHECK2-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK2-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK2-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64* // CHECK2-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK2-NEXT: store i64 [[TMP61]], i64* [[TMP69]], align 8 // CHECK2-NEXT: store i64 [[TMP61]], i64* [[TMP69]], align 8
// CHECK2-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK2-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK2-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64* // CHECK2-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64*
// CHECK2-NEXT: store i64 [[TMP61]], i64* [[TMP71]], align 8 // CHECK2-NEXT: store i64 [[TMP61]], i64* [[TMP71]], align 8
// CHECK2-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK2-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK2-NEXT: store i64 4, i64* [[TMP72]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP72]], align 8
// CHECK2-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK2-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP73]], align 8 // CHECK2-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [10 x float]**
// CHECK2-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP74]], align 8
// CHECK2-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to [10 x float]** // CHECK2-NEXT: [[TMP75:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP75]], align 8 // CHECK2-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [10 x float]**
// CHECK2-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP76]], align 8
// CHECK2-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to [10 x float]** // CHECK2-NEXT: [[TMP77:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP77]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP77]], align 8
// CHECK2-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK2-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK2-NEXT: store i64 40, i64* [[TMP78]], align 8 // CHECK2-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK2-NEXT: [[TMP79:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP79]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP79]], align 8 // CHECK2-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK2-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK2-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64* // CHECK2-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64*
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP81]], align 8 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP81]], align 8
// CHECK2-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK2-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK2-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i64* // CHECK2-NEXT: store i8* null, i8** [[TMP82]], align 8
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP83]], align 8 // CHECK2-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK2-NEXT: [[TMP84:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK2-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to float**
// CHECK2-NEXT: store i64 8, i64* [[TMP84]], align 8 // CHECK2-NEXT: store float* [[VLA]], float** [[TMP84]], align 8
// CHECK2-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK2-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK2-NEXT: store i8* null, i8** [[TMP85]], align 8 // CHECK2-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to float**
// CHECK2-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3 // CHECK2-NEXT: store float* [[VLA]], float** [[TMP86]], align 8
// CHECK2-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to float** // CHECK2-NEXT: store i64 [[TMP65]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK2-NEXT: store float* [[VLA]], float** [[TMP87]], align 8 // CHECK2-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK2-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3 // CHECK2-NEXT: store i8* null, i8** [[TMP87]], align 8
// CHECK2-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to float** // CHECK2-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4
// CHECK2-NEXT: store float* [[VLA]], float** [[TMP89]], align 8 // CHECK2-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to [5 x [10 x double]]**
// CHECK2-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP89]], align 8
// CHECK2-NEXT: store i64 [[TMP65]], i64* [[TMP90]], align 8 // CHECK2-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4
// CHECK2-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3 // CHECK2-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to [5 x [10 x double]]**
// CHECK2-NEXT: store i8* null, i8** [[TMP91]], align 8 // CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP91]], align 8
// CHECK2-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4 // CHECK2-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4
// CHECK2-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [5 x [10 x double]]** // CHECK2-NEXT: store i8* null, i8** [[TMP92]], align 8
// CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP93]], align 8 // CHECK2-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5
// CHECK2-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4 // CHECK2-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i64*
// CHECK2-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [5 x [10 x double]]** // CHECK2-NEXT: store i64 5, i64* [[TMP94]], align 8
// CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP95]], align 8 // CHECK2-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5
// CHECK2-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK2-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64*
// CHECK2-NEXT: store i64 400, i64* [[TMP96]], align 8 // CHECK2-NEXT: store i64 5, i64* [[TMP96]], align 8
// CHECK2-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4 // CHECK2-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5
// CHECK2-NEXT: store i8* null, i8** [[TMP97]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP97]], align 8
// CHECK2-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5 // CHECK2-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6
// CHECK2-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64* // CHECK2-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64*
// CHECK2-NEXT: store i64 5, i64* [[TMP99]], align 8 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP99]], align 8
// CHECK2-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5 // CHECK2-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6
// CHECK2-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64* // CHECK2-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64*
// CHECK2-NEXT: store i64 5, i64* [[TMP101]], align 8 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP101]], align 8
// CHECK2-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK2-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6
// CHECK2-NEXT: store i64 8, i64* [[TMP102]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK2-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5 // CHECK2-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7
// CHECK2-NEXT: store i8* null, i8** [[TMP103]], align 8 // CHECK2-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to double**
// CHECK2-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6 // CHECK2-NEXT: store double* [[VLA1]], double** [[TMP104]], align 8
// CHECK2-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to i64* // CHECK2-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7
// CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP105]], align 8 // CHECK2-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to double**
// CHECK2-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6 // CHECK2-NEXT: store double* [[VLA1]], double** [[TMP106]], align 8
// CHECK2-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to i64* // CHECK2-NEXT: store i64 [[TMP67]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP107]], align 8 // CHECK2-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7
// CHECK2-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK2-NEXT: store i8* null, i8** [[TMP107]], align 8
// CHECK2-NEXT: store i64 8, i64* [[TMP108]], align 8 // CHECK2-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8
// CHECK2-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6 // CHECK2-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to %struct.TT**
// CHECK2-NEXT: store i8* null, i8** [[TMP109]], align 8 // CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP109]], align 8
// CHECK2-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7 // CHECK2-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8
// CHECK2-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to double** // CHECK2-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to %struct.TT**
// CHECK2-NEXT: store double* [[VLA1]], double** [[TMP111]], align 8 // CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP111]], align 8
// CHECK2-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7 // CHECK2-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK2-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to double** // CHECK2-NEXT: store i8* null, i8** [[TMP112]], align 8
// CHECK2-NEXT: store double* [[VLA1]], double** [[TMP113]], align 8 // CHECK2-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK2-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK2-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to i64*
// CHECK2-NEXT: store i64 [[TMP67]], i64* [[TMP114]], align 8 // CHECK2-NEXT: store i64 [[TMP63]], i64* [[TMP114]], align 8
// CHECK2-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7 // CHECK2-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK2-NEXT: store i8* null, i8** [[TMP115]], align 8 // CHECK2-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to i64*
// CHECK2-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8 // CHECK2-NEXT: store i64 [[TMP63]], i64* [[TMP116]], align 8
// CHECK2-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to %struct.TT** // CHECK2-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP117]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP117]], align 8
// CHECK2-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8 // CHECK2-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK2-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to %struct.TT** // CHECK2-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP119]], align 8 // CHECK2-NEXT: [[TMP120:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP118]], i8** [[TMP119]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK2-NEXT: [[TMP121:%.*]] = icmp ne i32 [[TMP120]], 0
// CHECK2-NEXT: store i64 16, i64* [[TMP120]], align 8 // CHECK2-NEXT: br i1 [[TMP121]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK2-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK2-NEXT: store i8* null, i8** [[TMP121]], align 8
// CHECK2-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK2-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i64*
// CHECK2-NEXT: store i64 [[TMP63]], i64* [[TMP123]], align 8
// CHECK2-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK2-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to i64*
// CHECK2-NEXT: store i64 [[TMP63]], i64* [[TMP125]], align 8
// CHECK2-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK2-NEXT: store i64 4, i64* [[TMP126]], align 8
// CHECK2-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK2-NEXT: store i8* null, i8** [[TMP127]], align 8
// CHECK2-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK2-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK2-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK2-NEXT: [[TMP131:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP128]], i8** [[TMP129]], i64* [[TMP130]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP132:%.*]] = icmp ne i32 [[TMP131]], 0
// CHECK2-NEXT: br i1 [[TMP132]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK2: omp_offload.failed23: // CHECK2: omp_offload.failed23:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT24]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT24]]
// CHECK2: omp_offload.cont24: // CHECK2: omp_offload.cont24:
// CHECK2-NEXT: br label [[OMP_IF_END26:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END26:%.*]]
// CHECK2: omp_if.else25: // CHECK2: omp_if.else25:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_IF_END26]] // CHECK2-NEXT: br label [[OMP_IF_END26]]
// CHECK2: omp_if.end26: // CHECK2: omp_if.end26:
// CHECK2-NEXT: [[TMP133:%.*]] = load i32, i32* [[A]], align 4 // CHECK2-NEXT: [[TMP122:%.*]] = load i32, i32* [[A]], align 4
// CHECK2-NEXT: [[TMP134:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: [[TMP123:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP134]]) // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP123]])
// CHECK2-NEXT: ret i32 [[TMP133]] // CHECK2-NEXT: ret i32 [[TMP122]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96
// CHECK2-SAME: () #[[ATTR2:[0-9]+]] { // CHECK2-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
▲ Show 20 Lines • Show All 711 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK2-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK2-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK2-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK2-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 14 Lines
// CHECK2-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK2-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK2-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK2-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK2-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK2-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK2-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK2-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK2-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK2-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK2-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK2-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK2-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK2-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK2-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK2-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK2-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK2-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK2-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK2-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK2-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK2-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK2-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK2-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK2-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK2-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK2-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK2-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK2-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK2-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8 // CHECK2-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK2-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK2-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK2-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK2-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK2-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK2-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK2-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_IF_END]] // CHECK2-NEXT: br label [[OMP_IF_END]]
// CHECK2: omp_if.end: // CHECK2: omp_if.end:
// CHECK2-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK2-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK2-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK2-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK2-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK2-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK2-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK2-NEXT: ret i32 [[ADD4]] // CHECK2-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK2-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK2-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK2-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK2-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK2-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK2-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK2-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK2-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK2-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK2-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
Show All 22 Lines
// CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 109 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK2-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK2-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK2-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK2-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK2-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK2-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK2-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
Show All 29 Lines
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 117 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[A]], align 4 // CHECK3-NEXT: store i32 0, i32* [[A]], align 4
// CHECK3-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK3-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK3-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK3-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP64:%.*]] = mul nuw i32 [[TMP63]], 8 // CHECK3-NEXT: [[TMP64:%.*]] = mul nuw i32 [[TMP63]], 8
// CHECK3-NEXT: [[TMP65:%.*]] = sext i32 [[TMP64]] to i64 // CHECK3-NEXT: [[TMP65:%.*]] = sext i32 [[TMP64]] to i64
// CHECK3-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0 // CHECK3-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK3-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32* // CHECK3-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32*
// CHECK3-NEXT: store i32 [[TMP57]], i32* [[TMP67]], align 4 // CHECK3-NEXT: store i32 [[TMP57]], i32* [[TMP67]], align 4
// CHECK3-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0 // CHECK3-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK3-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32* // CHECK3-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK3-NEXT: store i32 [[TMP57]], i32* [[TMP69]], align 4 // CHECK3-NEXT: store i32 [[TMP57]], i32* [[TMP69]], align 4
// CHECK3-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK3-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0
// CHECK3-NEXT: store i64 4, i64* [[TMP70]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP70]], align 4
// CHECK3-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0 // CHECK3-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP71]], align 4 // CHECK3-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [10 x float]**
// CHECK3-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1 // CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP72]], align 4
// CHECK3-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to [10 x float]** // CHECK3-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1
// CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP73]], align 4 // CHECK3-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [10 x float]**
// CHECK3-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1 // CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP74]], align 4
// CHECK3-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to [10 x float]** // CHECK3-NEXT: [[TMP75:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1
// CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP75]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP75]], align 4
// CHECK3-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK3-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK3-NEXT: store i64 40, i64* [[TMP76]], align 4 // CHECK3-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32*
// CHECK3-NEXT: [[TMP77:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP77]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP77]], align 4 // CHECK3-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2
// CHECK3-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK3-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32* // CHECK3-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP79]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP79]], align 4
// CHECK3-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2 // CHECK3-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2
// CHECK3-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32* // CHECK3-NEXT: store i8* null, i8** [[TMP80]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP81]], align 4 // CHECK3-NEXT: [[TMP81:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3
// CHECK3-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK3-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to float**
// CHECK3-NEXT: store i64 4, i64* [[TMP82]], align 4 // CHECK3-NEXT: store float* [[VLA]], float** [[TMP82]], align 4
// CHECK3-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2 // CHECK3-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP83]], align 4 // CHECK3-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to float**
// CHECK3-NEXT: [[TMP84:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3 // CHECK3-NEXT: store float* [[VLA]], float** [[TMP84]], align 4
// CHECK3-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to float** // CHECK3-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK3-NEXT: store float* [[VLA]], float** [[TMP85]], align 4 // CHECK3-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3
// CHECK3-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3 // CHECK3-NEXT: store i8* null, i8** [[TMP85]], align 4
// CHECK3-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to float** // CHECK3-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4
// CHECK3-NEXT: store float* [[VLA]], float** [[TMP87]], align 4 // CHECK3-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to [5 x [10 x double]]**
// CHECK3-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP87]], align 4
// CHECK3-NEXT: store i64 [[TMP62]], i64* [[TMP88]], align 4 // CHECK3-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4
// CHECK3-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3 // CHECK3-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to [5 x [10 x double]]**
// CHECK3-NEXT: store i8* null, i8** [[TMP89]], align 4 // CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP89]], align 4
// CHECK3-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4 // CHECK3-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4
// CHECK3-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to [5 x [10 x double]]** // CHECK3-NEXT: store i8* null, i8** [[TMP90]], align 4
// CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP91]], align 4 // CHECK3-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5
// CHECK3-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4 // CHECK3-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32*
// CHECK3-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [5 x [10 x double]]** // CHECK3-NEXT: store i32 5, i32* [[TMP92]], align 4
// CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP93]], align 4 // CHECK3-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5
// CHECK3-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK3-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i32*
// CHECK3-NEXT: store i64 400, i64* [[TMP94]], align 4 // CHECK3-NEXT: store i32 5, i32* [[TMP94]], align 4
// CHECK3-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4 // CHECK3-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5
// CHECK3-NEXT: store i8* null, i8** [[TMP95]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP95]], align 4
// CHECK3-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5 // CHECK3-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6
// CHECK3-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i32* // CHECK3-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i32*
// CHECK3-NEXT: store i32 5, i32* [[TMP97]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP97]], align 4
// CHECK3-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5 // CHECK3-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6
// CHECK3-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32* // CHECK3-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32*
// CHECK3-NEXT: store i32 5, i32* [[TMP99]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP99]], align 4
// CHECK3-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK3-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6
// CHECK3-NEXT: store i64 4, i64* [[TMP100]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP100]], align 4
// CHECK3-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5 // CHECK3-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7
// CHECK3-NEXT: store i8* null, i8** [[TMP101]], align 4 // CHECK3-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to double**
// CHECK3-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6 // CHECK3-NEXT: store double* [[VLA1]], double** [[TMP102]], align 4
// CHECK3-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to i32* // CHECK3-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7
// CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP103]], align 4 // CHECK3-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to double**
// CHECK3-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6 // CHECK3-NEXT: store double* [[VLA1]], double** [[TMP104]], align 4
// CHECK3-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to i32* // CHECK3-NEXT: store i64 [[TMP65]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP105]], align 4 // CHECK3-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7
// CHECK3-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK3-NEXT: store i8* null, i8** [[TMP105]], align 4
// CHECK3-NEXT: store i64 4, i64* [[TMP106]], align 4 // CHECK3-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8
// CHECK3-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6 // CHECK3-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to %struct.TT**
// CHECK3-NEXT: store i8* null, i8** [[TMP107]], align 4 // CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP107]], align 4
// CHECK3-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7 // CHECK3-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8
// CHECK3-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to double** // CHECK3-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to %struct.TT**
// CHECK3-NEXT: store double* [[VLA1]], double** [[TMP109]], align 4 // CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP109]], align 4
// CHECK3-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7 // CHECK3-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK3-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to double** // CHECK3-NEXT: store i8* null, i8** [[TMP110]], align 4
// CHECK3-NEXT: store double* [[VLA1]], double** [[TMP111]], align 4 // CHECK3-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK3-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK3-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to i32*
// CHECK3-NEXT: store i64 [[TMP65]], i64* [[TMP112]], align 4 // CHECK3-NEXT: store i32 [[TMP59]], i32* [[TMP112]], align 4
// CHECK3-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7 // CHECK3-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK3-NEXT: store i8* null, i8** [[TMP113]], align 4 // CHECK3-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to i32*
// CHECK3-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8 // CHECK3-NEXT: store i32 [[TMP59]], i32* [[TMP114]], align 4
// CHECK3-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to %struct.TT** // CHECK3-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP115]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP115]], align 4
// CHECK3-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8 // CHECK3-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK3-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to %struct.TT** // CHECK3-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP117]], align 4 // CHECK3-NEXT: [[TMP118:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP116]], i8** [[TMP117]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK3-NEXT: [[TMP119:%.*]] = icmp ne i32 [[TMP118]], 0
// CHECK3-NEXT: store i64 12, i64* [[TMP118]], align 4 // CHECK3-NEXT: br i1 [[TMP119]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK3-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK3-NEXT: store i8* null, i8** [[TMP119]], align 4
// CHECK3-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK3-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i32*
// CHECK3-NEXT: store i32 [[TMP59]], i32* [[TMP121]], align 4
// CHECK3-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK3-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i32*
// CHECK3-NEXT: store i32 [[TMP59]], i32* [[TMP123]], align 4
// CHECK3-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK3-NEXT: store i64 4, i64* [[TMP124]], align 4
// CHECK3-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK3-NEXT: store i8* null, i8** [[TMP125]], align 4
// CHECK3-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK3-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK3-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK3-NEXT: [[TMP129:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP126]], i8** [[TMP127]], i64* [[TMP128]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP130:%.*]] = icmp ne i32 [[TMP129]], 0
// CHECK3-NEXT: br i1 [[TMP130]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK3: omp_offload.failed17: // CHECK3: omp_offload.failed17:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT18]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK3: omp_offload.cont18: // CHECK3: omp_offload.cont18:
// CHECK3-NEXT: br label [[OMP_IF_END20:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END20:%.*]]
// CHECK3: omp_if.else19: // CHECK3: omp_if.else19:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_IF_END20]] // CHECK3-NEXT: br label [[OMP_IF_END20]]
// CHECK3: omp_if.end20: // CHECK3: omp_if.end20:
// CHECK3-NEXT: [[TMP131:%.*]] = load i32, i32* [[A]], align 4 // CHECK3-NEXT: [[TMP120:%.*]] = load i32, i32* [[A]], align 4
// CHECK3-NEXT: [[TMP132:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: [[TMP121:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP132]]) // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP121]])
// CHECK3-NEXT: ret i32 [[TMP131]] // CHECK3-NEXT: ret i32 [[TMP120]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96
// CHECK3-SAME: () #[[ATTR2:[0-9]+]] { // CHECK3-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
▲ Show 20 Lines • Show All 691 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK3-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK3-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK3-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 13 Lines
// CHECK3-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK3-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK3-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK3-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK3-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK3-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK3-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK3-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK3-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK3-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK3-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK3-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK3-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK3-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK3-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK3-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK3-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK3-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK3-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK3-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK3-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK3-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK3-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4 // CHECK3-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK3-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK3-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK3-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK3-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_IF_END]] // CHECK3-NEXT: br label [[OMP_IF_END]]
// CHECK3: omp_if.end: // CHECK3: omp_if.end:
// CHECK3-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK3-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK3-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK3-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK3-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK3-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK3-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK3-NEXT: ret i32 [[ADD3]] // CHECK3-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK3-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK3-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK3-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK3-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK3-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
Show All 20 Lines
// CHECK3-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK3-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 106 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK3-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK3-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK3-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK3-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK3-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK3-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK3-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
Show All 26 Lines
// CHECK3-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK3-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK3-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 116 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK4-NEXT: store i32 0, i32* [[A]], align 4 // CHECK4-NEXT: store i32 0, i32* [[A]], align 4
// CHECK4-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK4-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK4-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK4-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP64:%.*]] = mul nuw i32 [[TMP63]], 8 // CHECK4-NEXT: [[TMP64:%.*]] = mul nuw i32 [[TMP63]], 8
// CHECK4-NEXT: [[TMP65:%.*]] = sext i32 [[TMP64]] to i64 // CHECK4-NEXT: [[TMP65:%.*]] = sext i32 [[TMP64]] to i64
// CHECK4-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0 // CHECK4-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK4-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32* // CHECK4-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32*
// CHECK4-NEXT: store i32 [[TMP57]], i32* [[TMP67]], align 4 // CHECK4-NEXT: store i32 [[TMP57]], i32* [[TMP67]], align 4
// CHECK4-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0 // CHECK4-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK4-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32* // CHECK4-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK4-NEXT: store i32 [[TMP57]], i32* [[TMP69]], align 4 // CHECK4-NEXT: store i32 [[TMP57]], i32* [[TMP69]], align 4
// CHECK4-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK4-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0
// CHECK4-NEXT: store i64 4, i64* [[TMP70]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP70]], align 4
// CHECK4-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0 // CHECK4-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1
// CHECK4-NEXT: store i8* null, i8** [[TMP71]], align 4 // CHECK4-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [10 x float]**
// CHECK4-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1 // CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP72]], align 4
// CHECK4-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to [10 x float]** // CHECK4-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1
// CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP73]], align 4 // CHECK4-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [10 x float]**
// CHECK4-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1 // CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP74]], align 4
// CHECK4-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to [10 x float]** // CHECK4-NEXT: [[TMP75:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1
// CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP75]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP75]], align 4
// CHECK4-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK4-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK4-NEXT: store i64 40, i64* [[TMP76]], align 4 // CHECK4-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32*
// CHECK4-NEXT: [[TMP77:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP77]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP77]], align 4 // CHECK4-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2
// CHECK4-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK4-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32* // CHECK4-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP79]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP79]], align 4
// CHECK4-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2 // CHECK4-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2
// CHECK4-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32* // CHECK4-NEXT: store i8* null, i8** [[TMP80]], align 4
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP81]], align 4 // CHECK4-NEXT: [[TMP81:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3
// CHECK4-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK4-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to float**
// CHECK4-NEXT: store i64 4, i64* [[TMP82]], align 4 // CHECK4-NEXT: store float* [[VLA]], float** [[TMP82]], align 4
// CHECK4-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2 // CHECK4-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP83]], align 4 // CHECK4-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to float**
// CHECK4-NEXT: [[TMP84:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3 // CHECK4-NEXT: store float* [[VLA]], float** [[TMP84]], align 4
// CHECK4-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to float** // CHECK4-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK4-NEXT: store float* [[VLA]], float** [[TMP85]], align 4 // CHECK4-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3
// CHECK4-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3 // CHECK4-NEXT: store i8* null, i8** [[TMP85]], align 4
// CHECK4-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to float** // CHECK4-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4
// CHECK4-NEXT: store float* [[VLA]], float** [[TMP87]], align 4 // CHECK4-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to [5 x [10 x double]]**
// CHECK4-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP87]], align 4
// CHECK4-NEXT: store i64 [[TMP62]], i64* [[TMP88]], align 4 // CHECK4-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4
// CHECK4-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3 // CHECK4-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to [5 x [10 x double]]**
// CHECK4-NEXT: store i8* null, i8** [[TMP89]], align 4 // CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP89]], align 4
// CHECK4-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4 // CHECK4-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4
// CHECK4-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to [5 x [10 x double]]** // CHECK4-NEXT: store i8* null, i8** [[TMP90]], align 4
// CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP91]], align 4 // CHECK4-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5
// CHECK4-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4 // CHECK4-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32*
// CHECK4-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [5 x [10 x double]]** // CHECK4-NEXT: store i32 5, i32* [[TMP92]], align 4
// CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP93]], align 4 // CHECK4-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5
// CHECK4-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK4-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i32*
// CHECK4-NEXT: store i64 400, i64* [[TMP94]], align 4 // CHECK4-NEXT: store i32 5, i32* [[TMP94]], align 4
// CHECK4-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4 // CHECK4-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5
// CHECK4-NEXT: store i8* null, i8** [[TMP95]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP95]], align 4
// CHECK4-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5 // CHECK4-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6
// CHECK4-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i32* // CHECK4-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i32*
// CHECK4-NEXT: store i32 5, i32* [[TMP97]], align 4 // CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP97]], align 4
// CHECK4-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5 // CHECK4-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6
// CHECK4-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32* // CHECK4-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32*
// CHECK4-NEXT: store i32 5, i32* [[TMP99]], align 4 // CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP99]], align 4
// CHECK4-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK4-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6
// CHECK4-NEXT: store i64 4, i64* [[TMP100]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP100]], align 4
// CHECK4-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5 // CHECK4-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7
// CHECK4-NEXT: store i8* null, i8** [[TMP101]], align 4 // CHECK4-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to double**
// CHECK4-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6 // CHECK4-NEXT: store double* [[VLA1]], double** [[TMP102]], align 4
// CHECK4-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to i32* // CHECK4-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7
// CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP103]], align 4 // CHECK4-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to double**
// CHECK4-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6 // CHECK4-NEXT: store double* [[VLA1]], double** [[TMP104]], align 4
// CHECK4-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to i32* // CHECK4-NEXT: store i64 [[TMP65]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP105]], align 4 // CHECK4-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7
// CHECK4-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK4-NEXT: store i8* null, i8** [[TMP105]], align 4
// CHECK4-NEXT: store i64 4, i64* [[TMP106]], align 4 // CHECK4-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8
// CHECK4-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6 // CHECK4-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to %struct.TT**
// CHECK4-NEXT: store i8* null, i8** [[TMP107]], align 4 // CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP107]], align 4
// CHECK4-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7 // CHECK4-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8
// CHECK4-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to double** // CHECK4-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to %struct.TT**
// CHECK4-NEXT: store double* [[VLA1]], double** [[TMP109]], align 4 // CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP109]], align 4
// CHECK4-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7 // CHECK4-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK4-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to double** // CHECK4-NEXT: store i8* null, i8** [[TMP110]], align 4
// CHECK4-NEXT: store double* [[VLA1]], double** [[TMP111]], align 4 // CHECK4-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK4-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK4-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to i32*
// CHECK4-NEXT: store i64 [[TMP65]], i64* [[TMP112]], align 4 // CHECK4-NEXT: store i32 [[TMP59]], i32* [[TMP112]], align 4
// CHECK4-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7 // CHECK4-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK4-NEXT: store i8* null, i8** [[TMP113]], align 4 // CHECK4-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to i32*
// CHECK4-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8 // CHECK4-NEXT: store i32 [[TMP59]], i32* [[TMP114]], align 4
// CHECK4-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to %struct.TT** // CHECK4-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP115]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP115]], align 4
// CHECK4-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8 // CHECK4-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK4-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to %struct.TT** // CHECK4-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP117]], align 4 // CHECK4-NEXT: [[TMP118:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP116]], i8** [[TMP117]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK4-NEXT: [[TMP119:%.*]] = icmp ne i32 [[TMP118]], 0
// CHECK4-NEXT: store i64 12, i64* [[TMP118]], align 4 // CHECK4-NEXT: br i1 [[TMP119]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK4-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK4-NEXT: store i8* null, i8** [[TMP119]], align 4
// CHECK4-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK4-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i32*
// CHECK4-NEXT: store i32 [[TMP59]], i32* [[TMP121]], align 4
// CHECK4-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK4-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i32*
// CHECK4-NEXT: store i32 [[TMP59]], i32* [[TMP123]], align 4
// CHECK4-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK4-NEXT: store i64 4, i64* [[TMP124]], align 4
// CHECK4-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK4-NEXT: store i8* null, i8** [[TMP125]], align 4
// CHECK4-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK4-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK4-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK4-NEXT: [[TMP129:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP126]], i8** [[TMP127]], i64* [[TMP128]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP130:%.*]] = icmp ne i32 [[TMP129]], 0
// CHECK4-NEXT: br i1 [[TMP130]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK4: omp_offload.failed17: // CHECK4: omp_offload.failed17:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT18]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK4: omp_offload.cont18: // CHECK4: omp_offload.cont18:
// CHECK4-NEXT: br label [[OMP_IF_END20:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END20:%.*]]
// CHECK4: omp_if.else19: // CHECK4: omp_if.else19:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_IF_END20]] // CHECK4-NEXT: br label [[OMP_IF_END20]]
// CHECK4: omp_if.end20: // CHECK4: omp_if.end20:
// CHECK4-NEXT: [[TMP131:%.*]] = load i32, i32* [[A]], align 4 // CHECK4-NEXT: [[TMP120:%.*]] = load i32, i32* [[A]], align 4
// CHECK4-NEXT: [[TMP132:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: [[TMP121:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP132]]) // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP121]])
// CHECK4-NEXT: ret i32 [[TMP131]] // CHECK4-NEXT: ret i32 [[TMP120]]
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96
// CHECK4-SAME: () #[[ATTR2:[0-9]+]] { // CHECK4-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
▲ Show 20 Lines • Show All 691 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK4-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK4-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK4-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 13 Lines
// CHECK4-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK4-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK4-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK4-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK4-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK4-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK4-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK4-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK4-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK4-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK4-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK4-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK4-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK4-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK4-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK4-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK4-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK4-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK4-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK4-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK4-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK4-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK4-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK4-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK4-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK4-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK4-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK4-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK4-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK4-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK4-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK4-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK4-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK4-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK4-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK4-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK4-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK4-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK4-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK4-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK4-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK4-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK4-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK4-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16** // CHECK4-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4 // CHECK4-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK4-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK4-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK4-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK4-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK4-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK4-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK4-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l216(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_IF_END]] // CHECK4-NEXT: br label [[OMP_IF_END]]
// CHECK4: omp_if.end: // CHECK4: omp_if.end:
// CHECK4-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK4-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK4-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK4-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK4-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK4-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK4-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK4-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK4-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK4-NEXT: ret i32 [[ADD3]] // CHECK4-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK4-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK4-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK4-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK4-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK4-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK4-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK4-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK4-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK4-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK4-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
Show All 20 Lines
// CHECK4-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK4-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK4-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK4-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 106 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK4-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK4-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK4-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK4-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK4-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK4-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK4-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK4-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
Show All 26 Lines
// CHECK4-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK4-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK4-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 117 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8
// CHECK5-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK5-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK5-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK5-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK5-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK5-NEXT: store i32 0, i32* [[A]], align 4 // CHECK5-NEXT: store i32 0, i32* [[A]], align 4
// CHECK5-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK5-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK5-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK5-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK5-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK5-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK5-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 124 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[TMP66:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK5-NEXT: [[TMP66:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK5-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP66]], 8 // CHECK5-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP66]], 8
// CHECK5-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK5-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK5-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64* // CHECK5-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK5-NEXT: store i64 [[TMP61]], i64* [[TMP69]], align 8 // CHECK5-NEXT: store i64 [[TMP61]], i64* [[TMP69]], align 8
// CHECK5-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK5-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK5-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64* // CHECK5-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64*
// CHECK5-NEXT: store i64 [[TMP61]], i64* [[TMP71]], align 8 // CHECK5-NEXT: store i64 [[TMP61]], i64* [[TMP71]], align 8
// CHECK5-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK5-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK5-NEXT: store i64 4, i64* [[TMP72]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP72]], align 8
// CHECK5-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK5-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK5-NEXT: store i8* null, i8** [[TMP73]], align 8 // CHECK5-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [10 x float]**
// CHECK5-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK5-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP74]], align 8
// CHECK5-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to [10 x float]** // CHECK5-NEXT: [[TMP75:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK5-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP75]], align 8 // CHECK5-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [10 x float]**
// CHECK5-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK5-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP76]], align 8
// CHECK5-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to [10 x float]** // CHECK5-NEXT: [[TMP77:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK5-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP77]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP77]], align 8
// CHECK5-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK5-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK5-NEXT: store i64 40, i64* [[TMP78]], align 8 // CHECK5-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK5-NEXT: [[TMP79:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP79]], align 8
// CHECK5-NEXT: store i8* null, i8** [[TMP79]], align 8 // CHECK5-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK5-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK5-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64* // CHECK5-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64*
// CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP81]], align 8 // CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP81]], align 8
// CHECK5-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK5-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK5-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i64* // CHECK5-NEXT: store i8* null, i8** [[TMP82]], align 8
// CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP83]], align 8 // CHECK5-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK5-NEXT: [[TMP84:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK5-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to float**
// CHECK5-NEXT: store i64 8, i64* [[TMP84]], align 8 // CHECK5-NEXT: store float* [[VLA]], float** [[TMP84]], align 8
// CHECK5-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK5-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK5-NEXT: store i8* null, i8** [[TMP85]], align 8 // CHECK5-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to float**
// CHECK5-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3 // CHECK5-NEXT: store float* [[VLA]], float** [[TMP86]], align 8
// CHECK5-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to float** // CHECK5-NEXT: store i64 [[TMP65]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK5-NEXT: store float* [[VLA]], float** [[TMP87]], align 8 // CHECK5-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK5-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3 // CHECK5-NEXT: store i8* null, i8** [[TMP87]], align 8
// CHECK5-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to float** // CHECK5-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4
// CHECK5-NEXT: store float* [[VLA]], float** [[TMP89]], align 8 // CHECK5-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to [5 x [10 x double]]**
// CHECK5-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK5-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP89]], align 8
// CHECK5-NEXT: store i64 [[TMP65]], i64* [[TMP90]], align 8 // CHECK5-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4
// CHECK5-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3 // CHECK5-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to [5 x [10 x double]]**
// CHECK5-NEXT: store i8* null, i8** [[TMP91]], align 8 // CHECK5-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP91]], align 8
// CHECK5-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4 // CHECK5-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4
// CHECK5-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [5 x [10 x double]]** // CHECK5-NEXT: store i8* null, i8** [[TMP92]], align 8
// CHECK5-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP93]], align 8 // CHECK5-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5
// CHECK5-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4 // CHECK5-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i64*
// CHECK5-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [5 x [10 x double]]** // CHECK5-NEXT: store i64 5, i64* [[TMP94]], align 8
// CHECK5-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP95]], align 8 // CHECK5-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5
// CHECK5-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK5-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64*
// CHECK5-NEXT: store i64 400, i64* [[TMP96]], align 8 // CHECK5-NEXT: store i64 5, i64* [[TMP96]], align 8
// CHECK5-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4 // CHECK5-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5
// CHECK5-NEXT: store i8* null, i8** [[TMP97]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP97]], align 8
// CHECK5-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5 // CHECK5-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6
// CHECK5-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64* // CHECK5-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64*
// CHECK5-NEXT: store i64 5, i64* [[TMP99]], align 8 // CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP99]], align 8
// CHECK5-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5 // CHECK5-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6
// CHECK5-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64* // CHECK5-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64*
// CHECK5-NEXT: store i64 5, i64* [[TMP101]], align 8 // CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP101]], align 8
// CHECK5-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK5-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6
// CHECK5-NEXT: store i64 8, i64* [[TMP102]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK5-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5 // CHECK5-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7
// CHECK5-NEXT: store i8* null, i8** [[TMP103]], align 8 // CHECK5-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to double**
// CHECK5-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6 // CHECK5-NEXT: store double* [[VLA1]], double** [[TMP104]], align 8
// CHECK5-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to i64* // CHECK5-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7
// CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP105]], align 8 // CHECK5-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to double**
// CHECK5-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6 // CHECK5-NEXT: store double* [[VLA1]], double** [[TMP106]], align 8
// CHECK5-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to i64* // CHECK5-NEXT: store i64 [[TMP67]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP107]], align 8 // CHECK5-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7
// CHECK5-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK5-NEXT: store i8* null, i8** [[TMP107]], align 8
// CHECK5-NEXT: store i64 8, i64* [[TMP108]], align 8 // CHECK5-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8
// CHECK5-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6 // CHECK5-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to %struct.TT**
// CHECK5-NEXT: store i8* null, i8** [[TMP109]], align 8 // CHECK5-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP109]], align 8
// CHECK5-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7 // CHECK5-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8
// CHECK5-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to double** // CHECK5-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to %struct.TT**
// CHECK5-NEXT: store double* [[VLA1]], double** [[TMP111]], align 8 // CHECK5-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP111]], align 8
// CHECK5-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7 // CHECK5-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK5-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to double** // CHECK5-NEXT: store i8* null, i8** [[TMP112]], align 8
// CHECK5-NEXT: store double* [[VLA1]], double** [[TMP113]], align 8 // CHECK5-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK5-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK5-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to i64*
// CHECK5-NEXT: store i64 [[TMP67]], i64* [[TMP114]], align 8 // CHECK5-NEXT: store i64 [[TMP63]], i64* [[TMP114]], align 8
// CHECK5-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7 // CHECK5-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK5-NEXT: store i8* null, i8** [[TMP115]], align 8 // CHECK5-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to i64*
// CHECK5-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8 // CHECK5-NEXT: store i64 [[TMP63]], i64* [[TMP116]], align 8
// CHECK5-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to %struct.TT** // CHECK5-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK5-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP117]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP117]], align 8
// CHECK5-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8 // CHECK5-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK5-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to %struct.TT** // CHECK5-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK5-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP119]], align 8 // CHECK5-NEXT: [[TMP120:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP118]], i8** [[TMP119]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK5-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK5-NEXT: [[TMP121:%.*]] = icmp ne i32 [[TMP120]], 0
// CHECK5-NEXT: store i64 16, i64* [[TMP120]], align 8 // CHECK5-NEXT: br i1 [[TMP121]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK5-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK5-NEXT: store i8* null, i8** [[TMP121]], align 8
// CHECK5-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK5-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i64*
// CHECK5-NEXT: store i64 [[TMP63]], i64* [[TMP123]], align 8
// CHECK5-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK5-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to i64*
// CHECK5-NEXT: store i64 [[TMP63]], i64* [[TMP125]], align 8
// CHECK5-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK5-NEXT: store i64 4, i64* [[TMP126]], align 8
// CHECK5-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK5-NEXT: store i8* null, i8** [[TMP127]], align 8
// CHECK5-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK5-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK5-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK5-NEXT: [[TMP131:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP128]], i8** [[TMP129]], i64* [[TMP130]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK5-NEXT: [[TMP132:%.*]] = icmp ne i32 [[TMP131]], 0
// CHECK5-NEXT: br i1 [[TMP132]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK5: omp_offload.failed23: // CHECK5: omp_offload.failed23:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT24]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT24]]
// CHECK5: omp_offload.cont24: // CHECK5: omp_offload.cont24:
// CHECK5-NEXT: br label [[OMP_IF_END26:%.*]] // CHECK5-NEXT: br label [[OMP_IF_END26:%.*]]
// CHECK5: omp_if.else25: // CHECK5: omp_if.else25:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_IF_END26]] // CHECK5-NEXT: br label [[OMP_IF_END26]]
// CHECK5: omp_if.end26: // CHECK5: omp_if.end26:
// CHECK5-NEXT: [[TMP133:%.*]] = load i32, i32* [[A]], align 4 // CHECK5-NEXT: [[TMP122:%.*]] = load i32, i32* [[A]], align 4
// CHECK5-NEXT: [[TMP134:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK5-NEXT: [[TMP123:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK5-NEXT: call void @llvm.stackrestore(i8* [[TMP134]]) // CHECK5-NEXT: call void @llvm.stackrestore(i8* [[TMP123]])
// CHECK5-NEXT: ret i32 [[TMP133]] // CHECK5-NEXT: ret i32 [[TMP122]]
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96
// CHECK5-SAME: () #[[ATTR2:[0-9]+]] { // CHECK5-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
▲ Show 20 Lines • Show All 713 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK5-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK5-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK5-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [6 x i64], align 8
// CHECK5-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK5-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK5-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK5-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK5-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK5-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK5-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK5-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK5-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK5-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 24 Lines
// CHECK5-NEXT: [[TMP11:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK5-NEXT: [[TMP11:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK5-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], 2 // CHECK5-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], 2
// CHECK5-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1** // CHECK5-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1**
// CHECK5-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 8 // CHECK5-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 8
// CHECK5-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double** // CHECK5-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double**
// CHECK5-NEXT: store double* [[A]], double** [[TMP16]], align 8 // CHECK5-NEXT: store double* [[A]], double** [[TMP16]], align 8
// CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK5-NEXT: store i64 8, i64* [[TMP17]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP17]], align 8
// CHECK5-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK5-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK5-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK5-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64*
// CHECK5-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK5-NEXT: store i64 [[TMP7]], i64* [[TMP19]], align 8
// CHECK5-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64* // CHECK5-NEXT: [[TMP20:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK5-NEXT: store i64 [[TMP7]], i64* [[TMP20]], align 8 // CHECK5-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK5-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK5-NEXT: store i64 [[TMP7]], i64* [[TMP21]], align 8
// CHECK5-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i64* // CHECK5-NEXT: [[TMP22:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK5-NEXT: store i64 [[TMP7]], i64* [[TMP22]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK5-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK5-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK5-NEXT: store i64 4, i64* [[TMP23]], align 8 // CHECK5-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i64*
// CHECK5-NEXT: [[TMP24:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK5-NEXT: store i64 2, i64* [[TMP24]], align 8
// CHECK5-NEXT: store i8* null, i8** [[TMP24]], align 8 // CHECK5-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK5-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK5-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64* // CHECK5-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK5-NEXT: store i64 2, i64* [[TMP26]], align 8 // CHECK5-NEXT: store i64 2, i64* [[TMP26]], align 8
// CHECK5-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK5-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK5-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64* // CHECK5-NEXT: store i8* null, i8** [[TMP27]], align 8
// CHECK5-NEXT: store i64 2, i64* [[TMP28]], align 8 // CHECK5-NEXT: [[TMP28:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK5-NEXT: [[TMP29:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK5-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64*
// CHECK5-NEXT: store i64 8, i64* [[TMP29]], align 8 // CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8
// CHECK5-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK5-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK5-NEXT: store i8* null, i8** [[TMP30]], align 8 // CHECK5-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64*
// CHECK5-NEXT: [[TMP31:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8
// CHECK5-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i64* // CHECK5-NEXT: [[TMP32:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP32]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP32]], align 8
// CHECK5-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK5-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK5-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i64* // CHECK5-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i16**
// CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP34]], align 8 // CHECK5-NEXT: store i16* [[VLA]], i16** [[TMP34]], align 8
// CHECK5-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK5-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK5-NEXT: store i64 8, i64* [[TMP35]], align 8 // CHECK5-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i16**
// CHECK5-NEXT: [[TMP36:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK5-NEXT: store i16* [[VLA]], i16** [[TMP36]], align 8
// CHECK5-NEXT: store i8* null, i8** [[TMP36]], align 8 // CHECK5-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK5-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK5-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK5-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i16** // CHECK5-NEXT: store i8* null, i8** [[TMP37]], align 8
// CHECK5-NEXT: store i16* [[VLA]], i16** [[TMP38]], align 8 // CHECK5-NEXT: [[TMP38:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK5-NEXT: [[TMP39:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK5-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64*
// CHECK5-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i16** // CHECK5-NEXT: store i64 [[TMP9]], i64* [[TMP39]], align 8
// CHECK5-NEXT: store i16* [[VLA]], i16** [[TMP40]], align 8 // CHECK5-NEXT: [[TMP40:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK5-NEXT: [[TMP41:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK5-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i64*
// CHECK5-NEXT: store i64 [[TMP12]], i64* [[TMP41]], align 8 // CHECK5-NEXT: store i64 [[TMP9]], i64* [[TMP41]], align 8
// CHECK5-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK5-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK5-NEXT: store i8* null, i8** [[TMP42]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP42]], align 8
// CHECK5-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK5-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i64* // CHECK5-NEXT: [[TMP44:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: store i64 [[TMP9]], i64* [[TMP44]], align 8 // CHECK5-NEXT: [[TMP45:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK5-NEXT: [[TMP45:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK5-NEXT: [[TOBOOL5:%.*]] = trunc i8 [[TMP45]] to i1
// CHECK5-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i64* // CHECK5-NEXT: [[TMP46:%.*]] = select i1 [[TOBOOL5]], i32 0, i32 1
// CHECK5-NEXT: store i64 [[TMP9]], i64* [[TMP46]], align 8 // CHECK5-NEXT: [[TMP47:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214.region_id, i32 6, i8** [[TMP43]], i8** [[TMP44]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 [[TMP46]])
// CHECK5-NEXT: [[TMP47:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK5-NEXT: [[TMP48:%.*]] = icmp ne i32 [[TMP47]], 0
// CHECK5-NEXT: store i64 1, i64* [[TMP47]], align 8 // CHECK5-NEXT: br i1 [[TMP48]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK5-NEXT: [[TMP48:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK5-NEXT: store i8* null, i8** [[TMP48]], align 8
// CHECK5-NEXT: [[TMP49:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP50:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP51:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK5-NEXT: [[TMP52:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK5-NEXT: [[TOBOOL5:%.*]] = trunc i8 [[TMP52]] to i1
// CHECK5-NEXT: [[TMP53:%.*]] = select i1 [[TOBOOL5]], i32 0, i32 1
// CHECK5-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214.region_id, i32 6, i8** [[TMP49]], i8** [[TMP50]], i64* [[TMP51]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 [[TMP53]])
// CHECK5-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK5-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK5: omp_offload.failed: // CHECK5: omp_offload.failed:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK5: omp_offload.cont: // CHECK5: omp_offload.cont:
// CHECK5-NEXT: br label [[OMP_IF_END:%.*]] // CHECK5-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK5: omp_if.else: // CHECK5: omp_if.else:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_IF_END]] // CHECK5-NEXT: br label [[OMP_IF_END]]
// CHECK5: omp_if.end: // CHECK5: omp_if.end:
// CHECK5-NEXT: [[TMP56:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK5-NEXT: [[TMP49:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP56]] // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP49]]
// CHECK5-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK5-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK5-NEXT: [[TMP57:%.*]] = load i16, i16* [[ARRAYIDX6]], align 2 // CHECK5-NEXT: [[TMP50:%.*]] = load i16, i16* [[ARRAYIDX6]], align 2
// CHECK5-NEXT: [[CONV7:%.*]] = sext i16 [[TMP57]] to i32 // CHECK5-NEXT: [[CONV7:%.*]] = sext i16 [[TMP50]] to i32
// CHECK5-NEXT: [[TMP58:%.*]] = load i32, i32* [[B]], align 4 // CHECK5-NEXT: [[TMP51:%.*]] = load i32, i32* [[B]], align 4
// CHECK5-NEXT: [[ADD8:%.*]] = add nsw i32 [[CONV7]], [[TMP58]] // CHECK5-NEXT: [[ADD8:%.*]] = add nsw i32 [[CONV7]], [[TMP51]]
// CHECK5-NEXT: [[TMP59:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK5-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK5-NEXT: call void @llvm.stackrestore(i8* [[TMP59]]) // CHECK5-NEXT: call void @llvm.stackrestore(i8* [[TMP52]])
// CHECK5-NEXT: ret i32 [[ADD8]] // CHECK5-NEXT: ret i32 [[ADD8]]
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK5-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK5-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK5-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK5-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK5-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK5-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK5-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK5-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK5-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK5-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK5-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK5-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK5-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK5-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK5-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK5-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK5-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK5-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK5-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK5: omp_offload.failed: // CHECK5: omp_offload.failed:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK5: omp_offload.cont: // CHECK5: omp_offload.cont:
// CHECK5-NEXT: br label [[OMP_IF_END:%.*]] // CHECK5-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK5: omp_if.else: // CHECK5: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK5-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK5-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK5-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK5-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK5-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK5-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK5-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK5-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK5-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK5-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK5-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK5-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK5-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK5-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK5: omp_offload.failed: // CHECK5: omp_offload.failed:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK5: omp_offload.cont: // CHECK5: omp_offload.cont:
// CHECK5-NEXT: br label [[OMP_IF_END:%.*]] // CHECK5-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK5: omp_if.else: // CHECK5: omp_if.else:
Show All 39 Lines
// CHECK5-NEXT: [[CONV5:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8* // CHECK5-NEXT: [[CONV5:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
// CHECK5-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 // CHECK5-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
// CHECK5-NEXT: store i8 [[FROMBOOL]], i8* [[CONV5]], align 1 // CHECK5-NEXT: store i8 [[FROMBOOL]], i8* [[CONV5]], align 1
// CHECK5-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK5-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK5-NEXT: [[TMP9:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK5-NEXT: [[TOBOOL6:%.*]] = trunc i8 [[TMP9]] to i1 // CHECK5-NEXT: [[TOBOOL6:%.*]] = trunc i8 [[TMP9]] to i1
// CHECK5-NEXT: br i1 [[TOBOOL6]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK5-NEXT: br i1 [[TOBOOL6]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
// CHECK5: omp_if.then: // CHECK5: omp_if.then:
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP1]], i64 [[TMP6]], i64 [[TMP2]], i64 [[TMP3]], i16* [[TMP4]], i64 [[TMP8]]) // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*, i64)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP1]], i64 [[TMP6]], i64 [[TMP2]], i64 [[TMP3]], i16* [[TMP4]], i64 [[TMP8]])
// CHECK5-NEXT: br label [[OMP_IF_END:%.*]] // CHECK5-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK5: omp_if.else: // CHECK5: omp_if.else:
// CHECK5-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]]) // CHECK5-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]])
// CHECK5-NEXT: store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4 // CHECK5-NEXT: store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK5-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
// CHECK5-NEXT: call void @.omp_outlined..9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]], %struct.S1* [[TMP1]], i64 [[TMP6]], i64 [[TMP2]], i64 [[TMP3]], i16* [[TMP4]], i64 [[TMP8]]) #[[ATTR4]] // CHECK5-NEXT: call void @.omp_outlined..10(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]], %struct.S1* [[TMP1]], i64 [[TMP6]], i64 [[TMP2]], i64 [[TMP3]], i16* [[TMP4]], i64 [[TMP8]]) #[[ATTR4]]
// CHECK5-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]]) // CHECK5-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]])
// CHECK5-NEXT: br label [[OMP_IF_END]] // CHECK5-NEXT: br label [[OMP_IF_END]]
// CHECK5: omp_if.end: // CHECK5: omp_if.end:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK5-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 172 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK5-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK5-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK5-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK5-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK5-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK5-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK5-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK5-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK5-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK5-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK5-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK5-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK5-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK5-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK5-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK5-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
Show All 29 Lines
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK5-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK5-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK5-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK5-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK5-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK5-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK5-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK5-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK5-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK5-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK5-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK5-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK5-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 118 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_PTRS11:%.*]] = alloca [2 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS12:%.*]] = alloca [2 x i8*], align 8
// CHECK6-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[A_CASTED15:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [10 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [10 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [10 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK6-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK6-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK6-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK6-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK6-NEXT: store i32 0, i32* [[A]], align 4 // CHECK6-NEXT: store i32 0, i32* [[A]], align 4
// CHECK6-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK6-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK6-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK6-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK6-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
// CHECK6-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8 // CHECK6-NEXT: store i8* [[TMP3]], i8** [[SAVED_STACK]], align 8
▲ Show 20 Lines • Show All 124 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[TMP66:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK6-NEXT: [[TMP66:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK6-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP66]], 8 // CHECK6-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP66]], 8
// CHECK6-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK6-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK6-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64* // CHECK6-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK6-NEXT: store i64 [[TMP61]], i64* [[TMP69]], align 8 // CHECK6-NEXT: store i64 [[TMP61]], i64* [[TMP69]], align 8
// CHECK6-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK6-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK6-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64* // CHECK6-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64*
// CHECK6-NEXT: store i64 [[TMP61]], i64* [[TMP71]], align 8 // CHECK6-NEXT: store i64 [[TMP61]], i64* [[TMP71]], align 8
// CHECK6-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK6-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK6-NEXT: store i64 4, i64* [[TMP72]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP72]], align 8
// CHECK6-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK6-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK6-NEXT: store i8* null, i8** [[TMP73]], align 8 // CHECK6-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [10 x float]**
// CHECK6-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK6-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP74]], align 8
// CHECK6-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to [10 x float]** // CHECK6-NEXT: [[TMP75:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK6-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP75]], align 8 // CHECK6-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to [10 x float]**
// CHECK6-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK6-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP76]], align 8
// CHECK6-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to [10 x float]** // CHECK6-NEXT: [[TMP77:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK6-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP77]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP77]], align 8
// CHECK6-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK6-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK6-NEXT: store i64 40, i64* [[TMP78]], align 8 // CHECK6-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK6-NEXT: [[TMP79:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP79]], align 8
// CHECK6-NEXT: store i8* null, i8** [[TMP79]], align 8 // CHECK6-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK6-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK6-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64* // CHECK6-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64*
// CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP81]], align 8 // CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP81]], align 8
// CHECK6-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK6-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK6-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i64* // CHECK6-NEXT: store i8* null, i8** [[TMP82]], align 8
// CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP83]], align 8 // CHECK6-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK6-NEXT: [[TMP84:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK6-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to float**
// CHECK6-NEXT: store i64 8, i64* [[TMP84]], align 8 // CHECK6-NEXT: store float* [[VLA]], float** [[TMP84]], align 8
// CHECK6-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK6-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK6-NEXT: store i8* null, i8** [[TMP85]], align 8 // CHECK6-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to float**
// CHECK6-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3 // CHECK6-NEXT: store float* [[VLA]], float** [[TMP86]], align 8
// CHECK6-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to float** // CHECK6-NEXT: store i64 [[TMP65]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK6-NEXT: store float* [[VLA]], float** [[TMP87]], align 8 // CHECK6-NEXT: [[TMP87:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK6-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3 // CHECK6-NEXT: store i8* null, i8** [[TMP87]], align 8
// CHECK6-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to float** // CHECK6-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4
// CHECK6-NEXT: store float* [[VLA]], float** [[TMP89]], align 8 // CHECK6-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to [5 x [10 x double]]**
// CHECK6-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK6-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP89]], align 8
// CHECK6-NEXT: store i64 [[TMP65]], i64* [[TMP90]], align 8 // CHECK6-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4
// CHECK6-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3 // CHECK6-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to [5 x [10 x double]]**
// CHECK6-NEXT: store i8* null, i8** [[TMP91]], align 8 // CHECK6-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP91]], align 8
// CHECK6-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 4 // CHECK6-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4
// CHECK6-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [5 x [10 x double]]** // CHECK6-NEXT: store i8* null, i8** [[TMP92]], align 8
// CHECK6-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP93]], align 8 // CHECK6-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5
// CHECK6-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 4 // CHECK6-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i64*
// CHECK6-NEXT: [[TMP95:%.*]] = bitcast i8** [[TMP94]] to [5 x [10 x double]]** // CHECK6-NEXT: store i64 5, i64* [[TMP94]], align 8
// CHECK6-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP95]], align 8 // CHECK6-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5
// CHECK6-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK6-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64*
// CHECK6-NEXT: store i64 400, i64* [[TMP96]], align 8 // CHECK6-NEXT: store i64 5, i64* [[TMP96]], align 8
// CHECK6-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 4 // CHECK6-NEXT: [[TMP97:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5
// CHECK6-NEXT: store i8* null, i8** [[TMP97]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP97]], align 8
// CHECK6-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 5 // CHECK6-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6
// CHECK6-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64* // CHECK6-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64*
// CHECK6-NEXT: store i64 5, i64* [[TMP99]], align 8 // CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP99]], align 8
// CHECK6-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 5 // CHECK6-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6
// CHECK6-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64* // CHECK6-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64*
// CHECK6-NEXT: store i64 5, i64* [[TMP101]], align 8 // CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP101]], align 8
// CHECK6-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK6-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6
// CHECK6-NEXT: store i64 8, i64* [[TMP102]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK6-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 5 // CHECK6-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7
// CHECK6-NEXT: store i8* null, i8** [[TMP103]], align 8 // CHECK6-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to double**
// CHECK6-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 6 // CHECK6-NEXT: store double* [[VLA1]], double** [[TMP104]], align 8
// CHECK6-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to i64* // CHECK6-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7
// CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP105]], align 8 // CHECK6-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to double**
// CHECK6-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 6 // CHECK6-NEXT: store double* [[VLA1]], double** [[TMP106]], align 8
// CHECK6-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to i64* // CHECK6-NEXT: store i64 [[TMP67]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 8
// CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP107]], align 8 // CHECK6-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7
// CHECK6-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK6-NEXT: store i8* null, i8** [[TMP107]], align 8
// CHECK6-NEXT: store i64 8, i64* [[TMP108]], align 8 // CHECK6-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8
// CHECK6-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 6 // CHECK6-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to %struct.TT**
// CHECK6-NEXT: store i8* null, i8** [[TMP109]], align 8 // CHECK6-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP109]], align 8
// CHECK6-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 7 // CHECK6-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8
// CHECK6-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to double** // CHECK6-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to %struct.TT**
// CHECK6-NEXT: store double* [[VLA1]], double** [[TMP111]], align 8 // CHECK6-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP111]], align 8
// CHECK6-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 7 // CHECK6-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK6-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to double** // CHECK6-NEXT: store i8* null, i8** [[TMP112]], align 8
// CHECK6-NEXT: store double* [[VLA1]], double** [[TMP113]], align 8 // CHECK6-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK6-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK6-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to i64*
// CHECK6-NEXT: store i64 [[TMP67]], i64* [[TMP114]], align 8 // CHECK6-NEXT: store i64 [[TMP63]], i64* [[TMP114]], align 8
// CHECK6-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 7 // CHECK6-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK6-NEXT: store i8* null, i8** [[TMP115]], align 8 // CHECK6-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to i64*
// CHECK6-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 8 // CHECK6-NEXT: store i64 [[TMP63]], i64* [[TMP116]], align 8
// CHECK6-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to %struct.TT** // CHECK6-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK6-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP117]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP117]], align 8
// CHECK6-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 8 // CHECK6-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK6-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to %struct.TT** // CHECK6-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK6-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP119]], align 8 // CHECK6-NEXT: [[TMP120:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP118]], i8** [[TMP119]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK6-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK6-NEXT: [[TMP121:%.*]] = icmp ne i32 [[TMP120]], 0
// CHECK6-NEXT: store i64 16, i64* [[TMP120]], align 8 // CHECK6-NEXT: br i1 [[TMP121]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK6-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 8
// CHECK6-NEXT: store i8* null, i8** [[TMP121]], align 8
// CHECK6-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 9
// CHECK6-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i64*
// CHECK6-NEXT: store i64 [[TMP63]], i64* [[TMP123]], align 8
// CHECK6-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 9
// CHECK6-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to i64*
// CHECK6-NEXT: store i64 [[TMP63]], i64* [[TMP125]], align 8
// CHECK6-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK6-NEXT: store i64 4, i64* [[TMP126]], align 8
// CHECK6-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 9
// CHECK6-NEXT: store i8* null, i8** [[TMP127]], align 8
// CHECK6-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK6-NEXT: [[TMP129:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK6-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK6-NEXT: [[TMP131:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP128]], i8** [[TMP129]], i64* [[TMP130]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK6-NEXT: [[TMP132:%.*]] = icmp ne i32 [[TMP131]], 0
// CHECK6-NEXT: br i1 [[TMP132]], label [[OMP_OFFLOAD_FAILED23:%.*]], label [[OMP_OFFLOAD_CONT24:%.*]]
// CHECK6: omp_offload.failed23: // CHECK6: omp_offload.failed23:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT24]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT24]]
// CHECK6: omp_offload.cont24: // CHECK6: omp_offload.cont24:
// CHECK6-NEXT: br label [[OMP_IF_END26:%.*]] // CHECK6-NEXT: br label [[OMP_IF_END26:%.*]]
// CHECK6: omp_if.else25: // CHECK6: omp_if.else25:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i64 [[TMP61]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP63]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_IF_END26]] // CHECK6-NEXT: br label [[OMP_IF_END26]]
// CHECK6: omp_if.end26: // CHECK6: omp_if.end26:
// CHECK6-NEXT: [[TMP133:%.*]] = load i32, i32* [[A]], align 4 // CHECK6-NEXT: [[TMP122:%.*]] = load i32, i32* [[A]], align 4
// CHECK6-NEXT: [[TMP134:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK6-NEXT: [[TMP123:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK6-NEXT: call void @llvm.stackrestore(i8* [[TMP134]]) // CHECK6-NEXT: call void @llvm.stackrestore(i8* [[TMP123]])
// CHECK6-NEXT: ret i32 [[TMP133]] // CHECK6-NEXT: ret i32 [[TMP122]]
// //
// //
// CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96 // CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96
// CHECK6-SAME: () #[[ATTR2:[0-9]+]] { // CHECK6-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK6-NEXT: entry: // CHECK6-NEXT: entry:
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK6-NEXT: ret void // CHECK6-NEXT: ret void
// //
▲ Show 20 Lines • Show All 713 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK6-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK6-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK6-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK6-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [6 x i64], align 8
// CHECK6-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK6-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK6-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK6-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK6-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK6-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK6-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK6-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK6-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK6-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
Show All 24 Lines
// CHECK6-NEXT: [[TMP11:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK6-NEXT: [[TMP11:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK6-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], 2 // CHECK6-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], 2
// CHECK6-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1** // CHECK6-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1**
// CHECK6-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 8 // CHECK6-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 8
// CHECK6-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double** // CHECK6-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double**
// CHECK6-NEXT: store double* [[A]], double** [[TMP16]], align 8 // CHECK6-NEXT: store double* [[A]], double** [[TMP16]], align 8
// CHECK6-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK6-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK6-NEXT: store i64 8, i64* [[TMP17]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP17]], align 8
// CHECK6-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK6-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK6-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK6-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64*
// CHECK6-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK6-NEXT: store i64 [[TMP7]], i64* [[TMP19]], align 8
// CHECK6-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64* // CHECK6-NEXT: [[TMP20:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK6-NEXT: store i64 [[TMP7]], i64* [[TMP20]], align 8 // CHECK6-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK6-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK6-NEXT: store i64 [[TMP7]], i64* [[TMP21]], align 8
// CHECK6-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i64* // CHECK6-NEXT: [[TMP22:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK6-NEXT: store i64 [[TMP7]], i64* [[TMP22]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK6-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK6-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK6-NEXT: store i64 4, i64* [[TMP23]], align 8 // CHECK6-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i64*
// CHECK6-NEXT: [[TMP24:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK6-NEXT: store i64 2, i64* [[TMP24]], align 8
// CHECK6-NEXT: store i8* null, i8** [[TMP24]], align 8 // CHECK6-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK6-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK6-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64* // CHECK6-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK6-NEXT: store i64 2, i64* [[TMP26]], align 8 // CHECK6-NEXT: store i64 2, i64* [[TMP26]], align 8
// CHECK6-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK6-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK6-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64* // CHECK6-NEXT: store i8* null, i8** [[TMP27]], align 8
// CHECK6-NEXT: store i64 2, i64* [[TMP28]], align 8 // CHECK6-NEXT: [[TMP28:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK6-NEXT: [[TMP29:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK6-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64*
// CHECK6-NEXT: store i64 8, i64* [[TMP29]], align 8 // CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8
// CHECK6-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK6-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK6-NEXT: store i8* null, i8** [[TMP30]], align 8 // CHECK6-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64*
// CHECK6-NEXT: [[TMP31:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8
// CHECK6-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i64* // CHECK6-NEXT: [[TMP32:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP32]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP32]], align 8
// CHECK6-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK6-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK6-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i64* // CHECK6-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i16**
// CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP34]], align 8 // CHECK6-NEXT: store i16* [[VLA]], i16** [[TMP34]], align 8
// CHECK6-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK6-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK6-NEXT: store i64 8, i64* [[TMP35]], align 8 // CHECK6-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i16**
// CHECK6-NEXT: [[TMP36:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK6-NEXT: store i16* [[VLA]], i16** [[TMP36]], align 8
// CHECK6-NEXT: store i8* null, i8** [[TMP36]], align 8 // CHECK6-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.11, i32 0, i32 4), align 8
// CHECK6-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK6-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK6-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i16** // CHECK6-NEXT: store i8* null, i8** [[TMP37]], align 8
// CHECK6-NEXT: store i16* [[VLA]], i16** [[TMP38]], align 8 // CHECK6-NEXT: [[TMP38:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK6-NEXT: [[TMP39:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK6-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64*
// CHECK6-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i16** // CHECK6-NEXT: store i64 [[TMP9]], i64* [[TMP39]], align 8
// CHECK6-NEXT: store i16* [[VLA]], i16** [[TMP40]], align 8 // CHECK6-NEXT: [[TMP40:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK6-NEXT: [[TMP41:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK6-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i64*
// CHECK6-NEXT: store i64 [[TMP12]], i64* [[TMP41]], align 8 // CHECK6-NEXT: store i64 [[TMP9]], i64* [[TMP41]], align 8
// CHECK6-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK6-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK6-NEXT: store i8* null, i8** [[TMP42]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP42]], align 8
// CHECK6-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK6-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i64* // CHECK6-NEXT: [[TMP44:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: store i64 [[TMP9]], i64* [[TMP44]], align 8 // CHECK6-NEXT: [[TMP45:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK6-NEXT: [[TMP45:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK6-NEXT: [[TOBOOL5:%.*]] = trunc i8 [[TMP45]] to i1
// CHECK6-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i64* // CHECK6-NEXT: [[TMP46:%.*]] = select i1 [[TOBOOL5]], i32 0, i32 1
// CHECK6-NEXT: store i64 [[TMP9]], i64* [[TMP46]], align 8 // CHECK6-NEXT: [[TMP47:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214.region_id, i32 6, i8** [[TMP43]], i8** [[TMP44]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 [[TMP46]])
// CHECK6-NEXT: [[TMP47:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK6-NEXT: [[TMP48:%.*]] = icmp ne i32 [[TMP47]], 0
// CHECK6-NEXT: store i64 1, i64* [[TMP47]], align 8 // CHECK6-NEXT: br i1 [[TMP48]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK6-NEXT: [[TMP48:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK6-NEXT: store i8* null, i8** [[TMP48]], align 8
// CHECK6-NEXT: [[TMP49:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP50:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP51:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK6-NEXT: [[TMP52:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK6-NEXT: [[TOBOOL5:%.*]] = trunc i8 [[TMP52]] to i1
// CHECK6-NEXT: [[TMP53:%.*]] = select i1 [[TOBOOL5]], i32 0, i32 1
// CHECK6-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214.region_id, i32 6, i8** [[TMP49]], i8** [[TMP50]], i64* [[TMP51]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 [[TMP53]])
// CHECK6-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK6-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK6: omp_offload.failed: // CHECK6: omp_offload.failed:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK6: omp_offload.cont: // CHECK6: omp_offload.cont:
// CHECK6-NEXT: br label [[OMP_IF_END:%.*]] // CHECK6-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK6: omp_if.else: // CHECK6: omp_if.else:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_IF_END]] // CHECK6-NEXT: br label [[OMP_IF_END]]
// CHECK6: omp_if.end: // CHECK6: omp_if.end:
// CHECK6-NEXT: [[TMP56:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK6-NEXT: [[TMP49:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK6-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP56]] // CHECK6-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP49]]
// CHECK6-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK6-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK6-NEXT: [[TMP57:%.*]] = load i16, i16* [[ARRAYIDX6]], align 2 // CHECK6-NEXT: [[TMP50:%.*]] = load i16, i16* [[ARRAYIDX6]], align 2
// CHECK6-NEXT: [[CONV7:%.*]] = sext i16 [[TMP57]] to i32 // CHECK6-NEXT: [[CONV7:%.*]] = sext i16 [[TMP50]] to i32
// CHECK6-NEXT: [[TMP58:%.*]] = load i32, i32* [[B]], align 4 // CHECK6-NEXT: [[TMP51:%.*]] = load i32, i32* [[B]], align 4
// CHECK6-NEXT: [[ADD8:%.*]] = add nsw i32 [[CONV7]], [[TMP58]] // CHECK6-NEXT: [[ADD8:%.*]] = add nsw i32 [[CONV7]], [[TMP51]]
// CHECK6-NEXT: [[TMP59:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK6-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK6-NEXT: call void @llvm.stackrestore(i8* [[TMP59]]) // CHECK6-NEXT: call void @llvm.stackrestore(i8* [[TMP52]])
// CHECK6-NEXT: ret i32 [[ADD8]] // CHECK6-NEXT: ret i32 [[ADD8]]
// //
// //
// CHECK6-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK6-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK6-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK6-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK6-NEXT: entry: // CHECK6-NEXT: entry:
// CHECK6-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8 // CHECK6-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 8
// CHECK6-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK6-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK6-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK6-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK6-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8 // CHECK6-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 8
// CHECK6-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK6-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK6-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK6-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK6-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK6-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK6-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK6-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK6-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK6: omp_offload.failed: // CHECK6: omp_offload.failed:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK6: omp_offload.cont: // CHECK6: omp_offload.cont:
// CHECK6-NEXT: br label [[OMP_IF_END:%.*]] // CHECK6-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK6: omp_if.else: // CHECK6: omp_if.else:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8 // CHECK6-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 8
// CHECK6-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK6-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK6-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK6-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK6-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK6-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK6-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK6-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK6-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK6-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK6-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK6-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK6-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK6-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK6-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK6: omp_offload.failed: // CHECK6: omp_offload.failed:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK6: omp_offload.cont: // CHECK6: omp_offload.cont:
// CHECK6-NEXT: br label [[OMP_IF_END:%.*]] // CHECK6-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK6: omp_if.else: // CHECK6: omp_if.else:
Show All 39 Lines
// CHECK6-NEXT: [[CONV5:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8* // CHECK6-NEXT: [[CONV5:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
// CHECK6-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 // CHECK6-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
// CHECK6-NEXT: store i8 [[FROMBOOL]], i8* [[CONV5]], align 1 // CHECK6-NEXT: store i8 [[FROMBOOL]], i8* [[CONV5]], align 1
// CHECK6-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK6-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK6-NEXT: [[TMP9:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK6-NEXT: [[TMP9:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK6-NEXT: [[TOBOOL6:%.*]] = trunc i8 [[TMP9]] to i1 // CHECK6-NEXT: [[TOBOOL6:%.*]] = trunc i8 [[TMP9]] to i1
// CHECK6-NEXT: br i1 [[TOBOOL6]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK6-NEXT: br i1 [[TOBOOL6]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
// CHECK6: omp_if.then: // CHECK6: omp_if.then:
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP1]], i64 [[TMP6]], i64 [[TMP2]], i64 [[TMP3]], i16* [[TMP4]], i64 [[TMP8]]) // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*, i64)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP1]], i64 [[TMP6]], i64 [[TMP2]], i64 [[TMP3]], i16* [[TMP4]], i64 [[TMP8]])
// CHECK6-NEXT: br label [[OMP_IF_END:%.*]] // CHECK6-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK6: omp_if.else: // CHECK6: omp_if.else:
// CHECK6-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]]) // CHECK6-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]])
// CHECK6-NEXT: store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4 // CHECK6-NEXT: store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK6-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
// CHECK6-NEXT: call void @.omp_outlined..9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]], %struct.S1* [[TMP1]], i64 [[TMP6]], i64 [[TMP2]], i64 [[TMP3]], i16* [[TMP4]], i64 [[TMP8]]) #[[ATTR4]] // CHECK6-NEXT: call void @.omp_outlined..10(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]], %struct.S1* [[TMP1]], i64 [[TMP6]], i64 [[TMP2]], i64 [[TMP3]], i16* [[TMP4]], i64 [[TMP8]]) #[[ATTR4]]
// CHECK6-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]]) // CHECK6-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]])
// CHECK6-NEXT: br label [[OMP_IF_END]] // CHECK6-NEXT: br label [[OMP_IF_END]]
// CHECK6: omp_if.end: // CHECK6: omp_if.end:
// CHECK6-NEXT: ret void // CHECK6-NEXT: ret void
// //
// //
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry: // CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK6-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK6-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 172 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK6-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK6-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK6-NEXT: [[CONV4:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK6-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2 // CHECK6-NEXT: store i16 [[TMP3]], i16* [[CONV4]], align 2
// CHECK6-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK6-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK6-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1 // CHECK6-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV2]], align 1
// CHECK6-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK6-NEXT: [[CONV5:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK6-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1 // CHECK6-NEXT: store i8 [[TMP5]], i8* [[CONV5]], align 1
// CHECK6-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK6-NEXT: [[TMP6:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK6-NEXT: ret void // CHECK6-NEXT: ret void
// //
// //
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry: // CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK6-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
Show All 29 Lines
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK6-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK6-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK6-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK6-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK6-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK6-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK6-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK6-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK6-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK6-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK6-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK6-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK6-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK6-NEXT: ret void // CHECK6-NEXT: ret void
// //
// //
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry: // CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK6-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 117 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4
// CHECK7-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK7-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK7-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK7-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK7-NEXT: store i32 0, i32* [[A]], align 4 // CHECK7-NEXT: store i32 0, i32* [[A]], align 4
// CHECK7-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK7-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK7-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK7-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK7-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK7-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK7-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK7-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[TMP64:%.*]] = mul nuw i32 [[TMP63]], 8 // CHECK7-NEXT: [[TMP64:%.*]] = mul nuw i32 [[TMP63]], 8
// CHECK7-NEXT: [[TMP65:%.*]] = sext i32 [[TMP64]] to i64 // CHECK7-NEXT: [[TMP65:%.*]] = sext i32 [[TMP64]] to i64
// CHECK7-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0 // CHECK7-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK7-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32* // CHECK7-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32*
// CHECK7-NEXT: store i32 [[TMP57]], i32* [[TMP67]], align 4 // CHECK7-NEXT: store i32 [[TMP57]], i32* [[TMP67]], align 4
// CHECK7-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0 // CHECK7-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK7-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32* // CHECK7-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK7-NEXT: store i32 [[TMP57]], i32* [[TMP69]], align 4 // CHECK7-NEXT: store i32 [[TMP57]], i32* [[TMP69]], align 4
// CHECK7-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK7-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0
// CHECK7-NEXT: store i64 4, i64* [[TMP70]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP70]], align 4
// CHECK7-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0 // CHECK7-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1
// CHECK7-NEXT: store i8* null, i8** [[TMP71]], align 4 // CHECK7-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [10 x float]**
// CHECK7-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1 // CHECK7-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP72]], align 4
// CHECK7-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to [10 x float]** // CHECK7-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1
// CHECK7-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP73]], align 4 // CHECK7-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [10 x float]**
// CHECK7-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1 // CHECK7-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP74]], align 4
// CHECK7-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to [10 x float]** // CHECK7-NEXT: [[TMP75:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1
// CHECK7-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP75]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP75]], align 4
// CHECK7-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK7-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK7-NEXT: store i64 40, i64* [[TMP76]], align 4 // CHECK7-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32*
// CHECK7-NEXT: [[TMP77:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP77]], align 4
// CHECK7-NEXT: store i8* null, i8** [[TMP77]], align 4 // CHECK7-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2
// CHECK7-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK7-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32* // CHECK7-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP79]], align 4 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP79]], align 4
// CHECK7-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2 // CHECK7-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2
// CHECK7-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32* // CHECK7-NEXT: store i8* null, i8** [[TMP80]], align 4
// CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP81]], align 4 // CHECK7-NEXT: [[TMP81:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3
// CHECK7-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK7-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to float**
// CHECK7-NEXT: store i64 4, i64* [[TMP82]], align 4 // CHECK7-NEXT: store float* [[VLA]], float** [[TMP82]], align 4
// CHECK7-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2 // CHECK7-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3
// CHECK7-NEXT: store i8* null, i8** [[TMP83]], align 4 // CHECK7-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to float**
// CHECK7-NEXT: [[TMP84:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3 // CHECK7-NEXT: store float* [[VLA]], float** [[TMP84]], align 4
// CHECK7-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to float** // CHECK7-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK7-NEXT: store float* [[VLA]], float** [[TMP85]], align 4 // CHECK7-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3
// CHECK7-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3 // CHECK7-NEXT: store i8* null, i8** [[TMP85]], align 4
// CHECK7-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to float** // CHECK7-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4
// CHECK7-NEXT: store float* [[VLA]], float** [[TMP87]], align 4 // CHECK7-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to [5 x [10 x double]]**
// CHECK7-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK7-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP87]], align 4
// CHECK7-NEXT: store i64 [[TMP62]], i64* [[TMP88]], align 4 // CHECK7-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4
// CHECK7-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3 // CHECK7-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to [5 x [10 x double]]**
// CHECK7-NEXT: store i8* null, i8** [[TMP89]], align 4 // CHECK7-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP89]], align 4
// CHECK7-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4 // CHECK7-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4
// CHECK7-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to [5 x [10 x double]]** // CHECK7-NEXT: store i8* null, i8** [[TMP90]], align 4
// CHECK7-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP91]], align 4 // CHECK7-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5
// CHECK7-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4 // CHECK7-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32*
// CHECK7-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [5 x [10 x double]]** // CHECK7-NEXT: store i32 5, i32* [[TMP92]], align 4
// CHECK7-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP93]], align 4 // CHECK7-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5
// CHECK7-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK7-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i32*
// CHECK7-NEXT: store i64 400, i64* [[TMP94]], align 4 // CHECK7-NEXT: store i32 5, i32* [[TMP94]], align 4
// CHECK7-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4 // CHECK7-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5
// CHECK7-NEXT: store i8* null, i8** [[TMP95]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP95]], align 4
// CHECK7-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5 // CHECK7-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6
// CHECK7-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i32* // CHECK7-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i32*
// CHECK7-NEXT: store i32 5, i32* [[TMP97]], align 4 // CHECK7-NEXT: store i32 [[TMP3]], i32* [[TMP97]], align 4
// CHECK7-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5 // CHECK7-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6
// CHECK7-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32* // CHECK7-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32*
// CHECK7-NEXT: store i32 5, i32* [[TMP99]], align 4 // CHECK7-NEXT: store i32 [[TMP3]], i32* [[TMP99]], align 4
// CHECK7-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK7-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6
// CHECK7-NEXT: store i64 4, i64* [[TMP100]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP100]], align 4
// CHECK7-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5 // CHECK7-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7
// CHECK7-NEXT: store i8* null, i8** [[TMP101]], align 4 // CHECK7-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to double**
// CHECK7-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6 // CHECK7-NEXT: store double* [[VLA1]], double** [[TMP102]], align 4
// CHECK7-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to i32* // CHECK7-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7
// CHECK7-NEXT: store i32 [[TMP3]], i32* [[TMP103]], align 4 // CHECK7-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to double**
// CHECK7-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6 // CHECK7-NEXT: store double* [[VLA1]], double** [[TMP104]], align 4
// CHECK7-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to i32* // CHECK7-NEXT: store i64 [[TMP65]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK7-NEXT: store i32 [[TMP3]], i32* [[TMP105]], align 4 // CHECK7-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7
// CHECK7-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK7-NEXT: store i8* null, i8** [[TMP105]], align 4
// CHECK7-NEXT: store i64 4, i64* [[TMP106]], align 4 // CHECK7-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8
// CHECK7-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6 // CHECK7-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to %struct.TT**
// CHECK7-NEXT: store i8* null, i8** [[TMP107]], align 4 // CHECK7-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP107]], align 4
// CHECK7-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7 // CHECK7-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8
// CHECK7-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to double** // CHECK7-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to %struct.TT**
// CHECK7-NEXT: store double* [[VLA1]], double** [[TMP109]], align 4 // CHECK7-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP109]], align 4
// CHECK7-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7 // CHECK7-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK7-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to double** // CHECK7-NEXT: store i8* null, i8** [[TMP110]], align 4
// CHECK7-NEXT: store double* [[VLA1]], double** [[TMP111]], align 4 // CHECK7-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK7-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK7-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to i32*
// CHECK7-NEXT: store i64 [[TMP65]], i64* [[TMP112]], align 4 // CHECK7-NEXT: store i32 [[TMP59]], i32* [[TMP112]], align 4
// CHECK7-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7 // CHECK7-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK7-NEXT: store i8* null, i8** [[TMP113]], align 4 // CHECK7-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to i32*
// CHECK7-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8 // CHECK7-NEXT: store i32 [[TMP59]], i32* [[TMP114]], align 4
// CHECK7-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to %struct.TT** // CHECK7-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK7-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP115]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP115]], align 4
// CHECK7-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8 // CHECK7-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK7-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to %struct.TT** // CHECK7-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK7-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP117]], align 4 // CHECK7-NEXT: [[TMP118:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP116]], i8** [[TMP117]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK7-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK7-NEXT: [[TMP119:%.*]] = icmp ne i32 [[TMP118]], 0
// CHECK7-NEXT: store i64 12, i64* [[TMP118]], align 4 // CHECK7-NEXT: br i1 [[TMP119]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK7-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK7-NEXT: store i8* null, i8** [[TMP119]], align 4
// CHECK7-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK7-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i32*
// CHECK7-NEXT: store i32 [[TMP59]], i32* [[TMP121]], align 4
// CHECK7-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK7-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i32*
// CHECK7-NEXT: store i32 [[TMP59]], i32* [[TMP123]], align 4
// CHECK7-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK7-NEXT: store i64 4, i64* [[TMP124]], align 4
// CHECK7-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK7-NEXT: store i8* null, i8** [[TMP125]], align 4
// CHECK7-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK7-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK7-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK7-NEXT: [[TMP129:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP126]], i8** [[TMP127]], i64* [[TMP128]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK7-NEXT: [[TMP130:%.*]] = icmp ne i32 [[TMP129]], 0
// CHECK7-NEXT: br i1 [[TMP130]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK7: omp_offload.failed17: // CHECK7: omp_offload.failed17:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT18]] // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK7: omp_offload.cont18: // CHECK7: omp_offload.cont18:
// CHECK7-NEXT: br label [[OMP_IF_END20:%.*]] // CHECK7-NEXT: br label [[OMP_IF_END20:%.*]]
// CHECK7: omp_if.else19: // CHECK7: omp_if.else19:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_IF_END20]] // CHECK7-NEXT: br label [[OMP_IF_END20]]
// CHECK7: omp_if.end20: // CHECK7: omp_if.end20:
// CHECK7-NEXT: [[TMP131:%.*]] = load i32, i32* [[A]], align 4 // CHECK7-NEXT: [[TMP120:%.*]] = load i32, i32* [[A]], align 4
// CHECK7-NEXT: [[TMP132:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK7-NEXT: [[TMP121:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK7-NEXT: call void @llvm.stackrestore(i8* [[TMP132]]) // CHECK7-NEXT: call void @llvm.stackrestore(i8* [[TMP121]])
// CHECK7-NEXT: ret i32 [[TMP131]] // CHECK7-NEXT: ret i32 [[TMP120]]
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96
// CHECK7-SAME: () #[[ATTR2:[0-9]+]] { // CHECK7-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
▲ Show 20 Lines • Show All 693 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK7-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK7-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK7-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [6 x i64], align 4
// CHECK7-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK7-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK7-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK7-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK7-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK7-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK7-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK7-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK7-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK7-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 23 Lines
// CHECK7-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 2 // CHECK7-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 2
// CHECK7-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64 // CHECK7-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64
// CHECK7-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1** // CHECK7-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1**
// CHECK7-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 4 // CHECK7-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 4
// CHECK7-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double** // CHECK7-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double**
// CHECK7-NEXT: store double* [[A]], double** [[TMP16]], align 4 // CHECK7-NEXT: store double* [[A]], double** [[TMP16]], align 4
// CHECK7-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK7-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK7-NEXT: store i64 8, i64* [[TMP17]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK7-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK7-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK7-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK7-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK7-NEXT: store i32 [[TMP6]], i32* [[TMP19]], align 4
// CHECK7-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK7-NEXT: [[TMP20:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK7-NEXT: store i32 [[TMP6]], i32* [[TMP20]], align 4 // CHECK7-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK7-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK7-NEXT: store i32 [[TMP6]], i32* [[TMP21]], align 4
// CHECK7-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32* // CHECK7-NEXT: [[TMP22:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK7-NEXT: store i32 [[TMP6]], i32* [[TMP22]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK7-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK7-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK7-NEXT: store i64 4, i64* [[TMP23]], align 4 // CHECK7-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK7-NEXT: [[TMP24:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK7-NEXT: store i32 2, i32* [[TMP24]], align 4
// CHECK7-NEXT: store i8* null, i8** [[TMP24]], align 4 // CHECK7-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK7-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK7-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK7-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK7-NEXT: store i32 2, i32* [[TMP26]], align 4 // CHECK7-NEXT: store i32 2, i32* [[TMP26]], align 4
// CHECK7-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK7-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK7-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32* // CHECK7-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK7-NEXT: store i32 2, i32* [[TMP28]], align 4 // CHECK7-NEXT: [[TMP28:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK7-NEXT: [[TMP29:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK7-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK7-NEXT: store i64 4, i64* [[TMP29]], align 4 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK7-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK7-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK7-NEXT: store i8* null, i8** [[TMP30]], align 4 // CHECK7-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32*
// CHECK7-NEXT: [[TMP31:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4
// CHECK7-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK7-NEXT: [[TMP32:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP32]], align 4
// CHECK7-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK7-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK7-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32* // CHECK7-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i16**
// CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP34]], align 4 // CHECK7-NEXT: store i16* [[VLA]], i16** [[TMP34]], align 4
// CHECK7-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK7-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK7-NEXT: store i64 4, i64* [[TMP35]], align 4 // CHECK7-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i16**
// CHECK7-NEXT: [[TMP36:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK7-NEXT: store i16* [[VLA]], i16** [[TMP36]], align 4
// CHECK7-NEXT: store i8* null, i8** [[TMP36]], align 4 // CHECK7-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK7-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK7-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK7-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i16** // CHECK7-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK7-NEXT: store i16* [[VLA]], i16** [[TMP38]], align 4 // CHECK7-NEXT: [[TMP38:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK7-NEXT: [[TMP39:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK7-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK7-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i16** // CHECK7-NEXT: store i32 [[TMP8]], i32* [[TMP39]], align 4
// CHECK7-NEXT: store i16* [[VLA]], i16** [[TMP40]], align 4 // CHECK7-NEXT: [[TMP40:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK7-NEXT: [[TMP41:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK7-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK7-NEXT: store i64 [[TMP12]], i64* [[TMP41]], align 4 // CHECK7-NEXT: store i32 [[TMP8]], i32* [[TMP41]], align 4
// CHECK7-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK7-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK7-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK7-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK7-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK7-NEXT: [[TMP44:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK7-NEXT: store i32 [[TMP8]], i32* [[TMP44]], align 4 // CHECK7-NEXT: [[TMP45:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK7-NEXT: [[TMP45:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK7-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP45]] to i1
// CHECK7-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK7-NEXT: [[TMP46:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
// CHECK7-NEXT: store i32 [[TMP8]], i32* [[TMP46]], align 4 // CHECK7-NEXT: [[TMP47:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214.region_id, i32 6, i8** [[TMP43]], i8** [[TMP44]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 [[TMP46]])
// CHECK7-NEXT: [[TMP47:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK7-NEXT: [[TMP48:%.*]] = icmp ne i32 [[TMP47]], 0
// CHECK7-NEXT: store i64 1, i64* [[TMP47]], align 4 // CHECK7-NEXT: br i1 [[TMP48]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK7-NEXT: [[TMP48:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK7-NEXT: store i8* null, i8** [[TMP48]], align 4
// CHECK7-NEXT: [[TMP49:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP50:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP51:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK7-NEXT: [[TMP52:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK7-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP52]] to i1
// CHECK7-NEXT: [[TMP53:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
// CHECK7-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214.region_id, i32 6, i8** [[TMP49]], i8** [[TMP50]], i64* [[TMP51]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 [[TMP53]])
// CHECK7-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK7-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK7: omp_offload.failed: // CHECK7: omp_offload.failed:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK7: omp_offload.cont: // CHECK7: omp_offload.cont:
// CHECK7-NEXT: br label [[OMP_IF_END:%.*]] // CHECK7-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK7: omp_if.else: // CHECK7: omp_if.else:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_IF_END]] // CHECK7-NEXT: br label [[OMP_IF_END]]
// CHECK7: omp_if.end: // CHECK7: omp_if.end:
// CHECK7-NEXT: [[TMP56:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK7-NEXT: [[TMP49:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP56]] // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP49]]
// CHECK7-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK7-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK7-NEXT: [[TMP57:%.*]] = load i16, i16* [[ARRAYIDX5]], align 2 // CHECK7-NEXT: [[TMP50:%.*]] = load i16, i16* [[ARRAYIDX5]], align 2
// CHECK7-NEXT: [[CONV6:%.*]] = sext i16 [[TMP57]] to i32 // CHECK7-NEXT: [[CONV6:%.*]] = sext i16 [[TMP50]] to i32
// CHECK7-NEXT: [[TMP58:%.*]] = load i32, i32* [[B]], align 4 // CHECK7-NEXT: [[TMP51:%.*]] = load i32, i32* [[B]], align 4
// CHECK7-NEXT: [[ADD7:%.*]] = add nsw i32 [[CONV6]], [[TMP58]] // CHECK7-NEXT: [[ADD7:%.*]] = add nsw i32 [[CONV6]], [[TMP51]]
// CHECK7-NEXT: [[TMP59:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK7-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK7-NEXT: call void @llvm.stackrestore(i8* [[TMP59]]) // CHECK7-NEXT: call void @llvm.stackrestore(i8* [[TMP52]])
// CHECK7-NEXT: ret i32 [[ADD7]] // CHECK7-NEXT: ret i32 [[ADD7]]
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK7-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK7-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK7-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK7-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK7-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK7-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK7-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK7-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK7-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK7-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK7-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK7-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK7-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK7-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK7-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK7-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK7-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK7-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK7-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK7: omp_offload.failed: // CHECK7: omp_offload.failed:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK7: omp_offload.cont: // CHECK7: omp_offload.cont:
// CHECK7-NEXT: br label [[OMP_IF_END:%.*]] // CHECK7-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK7: omp_if.else: // CHECK7: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK7-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK7-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK7-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK7-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK7-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK7-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK7-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK7-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK7-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK7-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK7-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK7-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK7-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK7-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK7-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK7-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK7: omp_offload.failed: // CHECK7: omp_offload.failed:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK7: omp_offload.cont: // CHECK7: omp_offload.cont:
// CHECK7-NEXT: br label [[OMP_IF_END:%.*]] // CHECK7-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK7: omp_if.else: // CHECK7: omp_if.else:
Show All 37 Lines
// CHECK7-NEXT: [[CONV3:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__CASTED]] to i8* // CHECK7-NEXT: [[CONV3:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__CASTED]] to i8*
// CHECK7-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 // CHECK7-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
// CHECK7-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1 // CHECK7-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK7-NEXT: [[TMP9:%.*]] = load i8, i8* [[CONV]], align 1 // CHECK7-NEXT: [[TMP9:%.*]] = load i8, i8* [[CONV]], align 1
// CHECK7-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP9]] to i1 // CHECK7-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP9]] to i1
// CHECK7-NEXT: br i1 [[TOBOOL4]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK7-NEXT: br i1 [[TOBOOL4]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
// CHECK7: omp_if.then: // CHECK7: omp_if.then:
// CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*, i32)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP1]], i32 [[TMP6]], i32 [[TMP2]], i32 [[TMP3]], i16* [[TMP4]], i32 [[TMP8]]) // CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*, i32)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP1]], i32 [[TMP6]], i32 [[TMP2]], i32 [[TMP3]], i16* [[TMP4]], i32 [[TMP8]])
// CHECK7-NEXT: br label [[OMP_IF_END:%.*]] // CHECK7-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK7: omp_if.else: // CHECK7: omp_if.else:
// CHECK7-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]]) // CHECK7-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]])
// CHECK7-NEXT: store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4 // CHECK7-NEXT: store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
// CHECK7-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK7-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
// CHECK7-NEXT: call void @.omp_outlined..9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]], %struct.S1* [[TMP1]], i32 [[TMP6]], i32 [[TMP2]], i32 [[TMP3]], i16* [[TMP4]], i32 [[TMP8]]) #[[ATTR4]] // CHECK7-NEXT: call void @.omp_outlined..10(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]], %struct.S1* [[TMP1]], i32 [[TMP6]], i32 [[TMP2]], i32 [[TMP3]], i16* [[TMP4]], i32 [[TMP8]]) #[[ATTR4]]
// CHECK7-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]]) // CHECK7-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]])
// CHECK7-NEXT: br label [[OMP_IF_END]] // CHECK7-NEXT: br label [[OMP_IF_END]]
// CHECK7: omp_if.end: // CHECK7: omp_if.end:
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK7-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 169 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK7-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK7-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK7-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK7-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK7-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK7-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK7-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK7-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK7-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK7-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK7-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK7-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
Show All 26 Lines
// CHECK7-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK7-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK7-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK7-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK7-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK7-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK7-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK7-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK7-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 116 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_PTRS7:%.*]] = alloca [2 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS8:%.*]] = alloca [2 x i8*], align 4
// CHECK8-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[A_CASTED11:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_BASEPTRS14:%.*]] = alloca [10 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_PTRS15:%.*]] = alloca [10 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS16:%.*]] = alloca [10 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK8-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK8-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK8-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK8-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK8-NEXT: store i32 0, i32* [[A]], align 4 // CHECK8-NEXT: store i32 0, i32* [[A]], align 4
// CHECK8-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK8-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK8-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK8-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK8-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK8-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
// CHECK8-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4 // CHECK8-NEXT: [[VLA:%.*]] = alloca float, i32 [[TMP1]], align 4
▲ Show 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[TMP64:%.*]] = mul nuw i32 [[TMP63]], 8 // CHECK8-NEXT: [[TMP64:%.*]] = mul nuw i32 [[TMP63]], 8
// CHECK8-NEXT: [[TMP65:%.*]] = sext i32 [[TMP64]] to i64 // CHECK8-NEXT: [[TMP65:%.*]] = sext i32 [[TMP64]] to i64
// CHECK8-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0 // CHECK8-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK8-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32* // CHECK8-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32*
// CHECK8-NEXT: store i32 [[TMP57]], i32* [[TMP67]], align 4 // CHECK8-NEXT: store i32 [[TMP57]], i32* [[TMP67]], align 4
// CHECK8-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0 // CHECK8-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK8-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32* // CHECK8-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK8-NEXT: store i32 [[TMP57]], i32* [[TMP69]], align 4 // CHECK8-NEXT: store i32 [[TMP57]], i32* [[TMP69]], align 4
// CHECK8-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK8-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0
// CHECK8-NEXT: store i64 4, i64* [[TMP70]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP70]], align 4
// CHECK8-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 0 // CHECK8-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1
// CHECK8-NEXT: store i8* null, i8** [[TMP71]], align 4 // CHECK8-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to [10 x float]**
// CHECK8-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 1 // CHECK8-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP72]], align 4
// CHECK8-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to [10 x float]** // CHECK8-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1
// CHECK8-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP73]], align 4 // CHECK8-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to [10 x float]**
// CHECK8-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 1 // CHECK8-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP74]], align 4
// CHECK8-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to [10 x float]** // CHECK8-NEXT: [[TMP75:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1
// CHECK8-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP75]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP75]], align 4
// CHECK8-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK8-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK8-NEXT: store i64 40, i64* [[TMP76]], align 4 // CHECK8-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32*
// CHECK8-NEXT: [[TMP77:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 1 // CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP77]], align 4
// CHECK8-NEXT: store i8* null, i8** [[TMP77]], align 4 // CHECK8-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2
// CHECK8-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 2
// CHECK8-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32* // CHECK8-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP79]], align 4 // CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP79]], align 4
// CHECK8-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 2 // CHECK8-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2
// CHECK8-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32* // CHECK8-NEXT: store i8* null, i8** [[TMP80]], align 4
// CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP81]], align 4 // CHECK8-NEXT: [[TMP81:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3
// CHECK8-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK8-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to float**
// CHECK8-NEXT: store i64 4, i64* [[TMP82]], align 4 // CHECK8-NEXT: store float* [[VLA]], float** [[TMP82]], align 4
// CHECK8-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 2 // CHECK8-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3
// CHECK8-NEXT: store i8* null, i8** [[TMP83]], align 4 // CHECK8-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to float**
// CHECK8-NEXT: [[TMP84:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 3 // CHECK8-NEXT: store float* [[VLA]], float** [[TMP84]], align 4
// CHECK8-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to float** // CHECK8-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK8-NEXT: store float* [[VLA]], float** [[TMP85]], align 4 // CHECK8-NEXT: [[TMP85:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3
// CHECK8-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 3 // CHECK8-NEXT: store i8* null, i8** [[TMP85]], align 4
// CHECK8-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to float** // CHECK8-NEXT: [[TMP86:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4
// CHECK8-NEXT: store float* [[VLA]], float** [[TMP87]], align 4 // CHECK8-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to [5 x [10 x double]]**
// CHECK8-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK8-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP87]], align 4
// CHECK8-NEXT: store i64 [[TMP62]], i64* [[TMP88]], align 4 // CHECK8-NEXT: [[TMP88:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4
// CHECK8-NEXT: [[TMP89:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 3 // CHECK8-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to [5 x [10 x double]]**
// CHECK8-NEXT: store i8* null, i8** [[TMP89]], align 4 // CHECK8-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP89]], align 4
// CHECK8-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 4 // CHECK8-NEXT: [[TMP90:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4
// CHECK8-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to [5 x [10 x double]]** // CHECK8-NEXT: store i8* null, i8** [[TMP90]], align 4
// CHECK8-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP91]], align 4 // CHECK8-NEXT: [[TMP91:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5
// CHECK8-NEXT: [[TMP92:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 4 // CHECK8-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32*
// CHECK8-NEXT: [[TMP93:%.*]] = bitcast i8** [[TMP92]] to [5 x [10 x double]]** // CHECK8-NEXT: store i32 5, i32* [[TMP92]], align 4
// CHECK8-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP93]], align 4 // CHECK8-NEXT: [[TMP93:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5
// CHECK8-NEXT: [[TMP94:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK8-NEXT: [[TMP94:%.*]] = bitcast i8** [[TMP93]] to i32*
// CHECK8-NEXT: store i64 400, i64* [[TMP94]], align 4 // CHECK8-NEXT: store i32 5, i32* [[TMP94]], align 4
// CHECK8-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 4 // CHECK8-NEXT: [[TMP95:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5
// CHECK8-NEXT: store i8* null, i8** [[TMP95]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP95]], align 4
// CHECK8-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 5 // CHECK8-NEXT: [[TMP96:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6
// CHECK8-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i32* // CHECK8-NEXT: [[TMP97:%.*]] = bitcast i8** [[TMP96]] to i32*
// CHECK8-NEXT: store i32 5, i32* [[TMP97]], align 4 // CHECK8-NEXT: store i32 [[TMP3]], i32* [[TMP97]], align 4
// CHECK8-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 5 // CHECK8-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6
// CHECK8-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32* // CHECK8-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32*
// CHECK8-NEXT: store i32 5, i32* [[TMP99]], align 4 // CHECK8-NEXT: store i32 [[TMP3]], i32* [[TMP99]], align 4
// CHECK8-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK8-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6
// CHECK8-NEXT: store i64 4, i64* [[TMP100]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP100]], align 4
// CHECK8-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 5 // CHECK8-NEXT: [[TMP101:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7
// CHECK8-NEXT: store i8* null, i8** [[TMP101]], align 4 // CHECK8-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to double**
// CHECK8-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 6 // CHECK8-NEXT: store double* [[VLA1]], double** [[TMP102]], align 4
// CHECK8-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to i32* // CHECK8-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7
// CHECK8-NEXT: store i32 [[TMP3]], i32* [[TMP103]], align 4 // CHECK8-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to double**
// CHECK8-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 6 // CHECK8-NEXT: store double* [[VLA1]], double** [[TMP104]], align 4
// CHECK8-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to i32* // CHECK8-NEXT: store i64 [[TMP65]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 7), align 4
// CHECK8-NEXT: store i32 [[TMP3]], i32* [[TMP105]], align 4 // CHECK8-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7
// CHECK8-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK8-NEXT: store i8* null, i8** [[TMP105]], align 4
// CHECK8-NEXT: store i64 4, i64* [[TMP106]], align 4 // CHECK8-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8
// CHECK8-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 6 // CHECK8-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to %struct.TT**
// CHECK8-NEXT: store i8* null, i8** [[TMP107]], align 4 // CHECK8-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP107]], align 4
// CHECK8-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 7 // CHECK8-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8
// CHECK8-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to double** // CHECK8-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to %struct.TT**
// CHECK8-NEXT: store double* [[VLA1]], double** [[TMP109]], align 4 // CHECK8-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP109]], align 4
// CHECK8-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 7 // CHECK8-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK8-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to double** // CHECK8-NEXT: store i8* null, i8** [[TMP110]], align 4
// CHECK8-NEXT: store double* [[VLA1]], double** [[TMP111]], align 4 // CHECK8-NEXT: [[TMP111:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK8-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK8-NEXT: [[TMP112:%.*]] = bitcast i8** [[TMP111]] to i32*
// CHECK8-NEXT: store i64 [[TMP65]], i64* [[TMP112]], align 4 // CHECK8-NEXT: store i32 [[TMP59]], i32* [[TMP112]], align 4
// CHECK8-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 7 // CHECK8-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK8-NEXT: store i8* null, i8** [[TMP113]], align 4 // CHECK8-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to i32*
// CHECK8-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 8 // CHECK8-NEXT: store i32 [[TMP59]], i32* [[TMP114]], align 4
// CHECK8-NEXT: [[TMP115:%.*]] = bitcast i8** [[TMP114]] to %struct.TT** // CHECK8-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK8-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP115]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP115]], align 4
// CHECK8-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 8 // CHECK8-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK8-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to %struct.TT** // CHECK8-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK8-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP117]], align 4 // CHECK8-NEXT: [[TMP118:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP116]], i8** [[TMP117]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK8-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK8-NEXT: [[TMP119:%.*]] = icmp ne i32 [[TMP118]], 0
// CHECK8-NEXT: store i64 12, i64* [[TMP118]], align 4 // CHECK8-NEXT: br i1 [[TMP119]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK8-NEXT: [[TMP119:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 8
// CHECK8-NEXT: store i8* null, i8** [[TMP119]], align 4
// CHECK8-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 9
// CHECK8-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i32*
// CHECK8-NEXT: store i32 [[TMP59]], i32* [[TMP121]], align 4
// CHECK8-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 9
// CHECK8-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i32*
// CHECK8-NEXT: store i32 [[TMP59]], i32* [[TMP123]], align 4
// CHECK8-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK8-NEXT: store i64 4, i64* [[TMP124]], align 4
// CHECK8-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS16]], i32 0, i32 9
// CHECK8-NEXT: store i8* null, i8** [[TMP125]], align 4
// CHECK8-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK8-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK8-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK8-NEXT: [[TMP129:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140.region_id, i32 10, i8** [[TMP126]], i8** [[TMP127]], i64* [[TMP128]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.8, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK8-NEXT: [[TMP130:%.*]] = icmp ne i32 [[TMP129]], 0
// CHECK8-NEXT: br i1 [[TMP130]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK8: omp_offload.failed17: // CHECK8: omp_offload.failed17:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT18]] // CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK8: omp_offload.cont18: // CHECK8: omp_offload.cont18:
// CHECK8-NEXT: br label [[OMP_IF_END20:%.*]] // CHECK8-NEXT: br label [[OMP_IF_END20:%.*]]
// CHECK8: omp_if.else19: // CHECK8: omp_if.else19:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l140(i32 [[TMP57]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP59]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_IF_END20]] // CHECK8-NEXT: br label [[OMP_IF_END20]]
// CHECK8: omp_if.end20: // CHECK8: omp_if.end20:
// CHECK8-NEXT: [[TMP131:%.*]] = load i32, i32* [[A]], align 4 // CHECK8-NEXT: [[TMP120:%.*]] = load i32, i32* [[A]], align 4
// CHECK8-NEXT: [[TMP132:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK8-NEXT: [[TMP121:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK8-NEXT: call void @llvm.stackrestore(i8* [[TMP132]]) // CHECK8-NEXT: call void @llvm.stackrestore(i8* [[TMP121]])
// CHECK8-NEXT: ret i32 [[TMP131]] // CHECK8-NEXT: ret i32 [[TMP120]]
// //
// //
// CHECK8-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96 // CHECK8-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l96
// CHECK8-SAME: () #[[ATTR2:[0-9]+]] { // CHECK8-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK8-NEXT: entry: // CHECK8-NEXT: entry:
// CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)) // CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK8-NEXT: ret void // CHECK8-NEXT: ret void
// //
▲ Show 20 Lines • Show All 693 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK8-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK8-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK8-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK8-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [6 x i64], align 4
// CHECK8-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK8-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK8-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK8-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK8-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK8-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK8-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK8-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK8-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK8-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK8-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK8-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK8-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK8-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
Show All 23 Lines
// CHECK8-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 2 // CHECK8-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 2
// CHECK8-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64 // CHECK8-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64
// CHECK8-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1** // CHECK8-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1**
// CHECK8-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 4 // CHECK8-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 4
// CHECK8-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double** // CHECK8-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double**
// CHECK8-NEXT: store double* [[A]], double** [[TMP16]], align 4 // CHECK8-NEXT: store double* [[A]], double** [[TMP16]], align 4
// CHECK8-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK8-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK8-NEXT: store i64 8, i64* [[TMP17]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK8-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK8-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK8-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK8-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK8-NEXT: store i32 [[TMP6]], i32* [[TMP19]], align 4
// CHECK8-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK8-NEXT: [[TMP20:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK8-NEXT: store i32 [[TMP6]], i32* [[TMP20]], align 4 // CHECK8-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK8-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK8-NEXT: store i32 [[TMP6]], i32* [[TMP21]], align 4
// CHECK8-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32* // CHECK8-NEXT: [[TMP22:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK8-NEXT: store i32 [[TMP6]], i32* [[TMP22]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK8-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK8-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK8-NEXT: store i64 4, i64* [[TMP23]], align 4 // CHECK8-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK8-NEXT: [[TMP24:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK8-NEXT: store i32 2, i32* [[TMP24]], align 4
// CHECK8-NEXT: store i8* null, i8** [[TMP24]], align 4 // CHECK8-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK8-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK8-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK8-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK8-NEXT: store i32 2, i32* [[TMP26]], align 4 // CHECK8-NEXT: store i32 2, i32* [[TMP26]], align 4
// CHECK8-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK8-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK8-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32* // CHECK8-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK8-NEXT: store i32 2, i32* [[TMP28]], align 4 // CHECK8-NEXT: [[TMP28:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK8-NEXT: [[TMP29:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK8-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK8-NEXT: store i64 4, i64* [[TMP29]], align 4 // CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK8-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK8-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK8-NEXT: store i8* null, i8** [[TMP30]], align 4 // CHECK8-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32*
// CHECK8-NEXT: [[TMP31:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4
// CHECK8-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK8-NEXT: [[TMP32:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP32]], align 4
// CHECK8-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK8-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK8-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32* // CHECK8-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i16**
// CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP34]], align 4 // CHECK8-NEXT: store i16* [[VLA]], i16** [[TMP34]], align 4
// CHECK8-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK8-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK8-NEXT: store i64 4, i64* [[TMP35]], align 4 // CHECK8-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i16**
// CHECK8-NEXT: [[TMP36:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK8-NEXT: store i16* [[VLA]], i16** [[TMP36]], align 4
// CHECK8-NEXT: store i8* null, i8** [[TMP36]], align 4 // CHECK8-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.11, i32 0, i32 4), align 4
// CHECK8-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK8-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK8-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i16** // CHECK8-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK8-NEXT: store i16* [[VLA]], i16** [[TMP38]], align 4 // CHECK8-NEXT: [[TMP38:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK8-NEXT: [[TMP39:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK8-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK8-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i16** // CHECK8-NEXT: store i32 [[TMP8]], i32* [[TMP39]], align 4
// CHECK8-NEXT: store i16* [[VLA]], i16** [[TMP40]], align 4 // CHECK8-NEXT: [[TMP40:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK8-NEXT: [[TMP41:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK8-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK8-NEXT: store i64 [[TMP12]], i64* [[TMP41]], align 4 // CHECK8-NEXT: store i32 [[TMP8]], i32* [[TMP41]], align 4
// CHECK8-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK8-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK8-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK8-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK8-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK8-NEXT: [[TMP44:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK8-NEXT: store i32 [[TMP8]], i32* [[TMP44]], align 4 // CHECK8-NEXT: [[TMP45:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK8-NEXT: [[TMP45:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK8-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP45]] to i1
// CHECK8-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK8-NEXT: [[TMP46:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
// CHECK8-NEXT: store i32 [[TMP8]], i32* [[TMP46]], align 4 // CHECK8-NEXT: [[TMP47:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214.region_id, i32 6, i8** [[TMP43]], i8** [[TMP44]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 1, i32 [[TMP46]])
// CHECK8-NEXT: [[TMP47:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK8-NEXT: [[TMP48:%.*]] = icmp ne i32 [[TMP47]], 0
// CHECK8-NEXT: store i64 1, i64* [[TMP47]], align 4 // CHECK8-NEXT: br i1 [[TMP48]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK8-NEXT: [[TMP48:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK8-NEXT: store i8* null, i8** [[TMP48]], align 4
// CHECK8-NEXT: [[TMP49:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP50:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP51:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK8-NEXT: [[TMP52:%.*]] = load i8, i8* [[DOTCAPTURE_EXPR_]], align 1
// CHECK8-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP52]] to i1
// CHECK8-NEXT: [[TMP53:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
// CHECK8-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214.region_id, i32 6, i8** [[TMP49]], i8** [[TMP50]], i64* [[TMP51]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 1, i32 [[TMP53]])
// CHECK8-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK8-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK8: omp_offload.failed: // CHECK8: omp_offload.failed:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK8: omp_offload.cont: // CHECK8: omp_offload.cont:
// CHECK8-NEXT: br label [[OMP_IF_END:%.*]] // CHECK8-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK8: omp_if.else: // CHECK8: omp_if.else:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l214(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_IF_END]] // CHECK8-NEXT: br label [[OMP_IF_END]]
// CHECK8: omp_if.end: // CHECK8: omp_if.end:
// CHECK8-NEXT: [[TMP56:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK8-NEXT: [[TMP49:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK8-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP56]] // CHECK8-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP49]]
// CHECK8-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK8-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK8-NEXT: [[TMP57:%.*]] = load i16, i16* [[ARRAYIDX5]], align 2 // CHECK8-NEXT: [[TMP50:%.*]] = load i16, i16* [[ARRAYIDX5]], align 2
// CHECK8-NEXT: [[CONV6:%.*]] = sext i16 [[TMP57]] to i32 // CHECK8-NEXT: [[CONV6:%.*]] = sext i16 [[TMP50]] to i32
// CHECK8-NEXT: [[TMP58:%.*]] = load i32, i32* [[B]], align 4 // CHECK8-NEXT: [[TMP51:%.*]] = load i32, i32* [[B]], align 4
// CHECK8-NEXT: [[ADD7:%.*]] = add nsw i32 [[CONV6]], [[TMP58]] // CHECK8-NEXT: [[ADD7:%.*]] = add nsw i32 [[CONV6]], [[TMP51]]
// CHECK8-NEXT: [[TMP59:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK8-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK8-NEXT: call void @llvm.stackrestore(i8* [[TMP59]]) // CHECK8-NEXT: call void @llvm.stackrestore(i8* [[TMP52]])
// CHECK8-NEXT: ret i32 [[ADD7]] // CHECK8-NEXT: ret i32 [[ADD7]]
// //
// //
// CHECK8-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK8-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK8-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK8-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK8-NEXT: entry: // CHECK8-NEXT: entry:
// CHECK8-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4 // CHECK8-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP23]], align 4
// CHECK8-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK8-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK8-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]** // CHECK8-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [10 x i32]**
// CHECK8-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4 // CHECK8-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP25]], align 4
// CHECK8-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK8-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK8-NEXT: store i8* null, i8** [[TMP26]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP26]], align 4
// CHECK8-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK8-NEXT: [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK8-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK8-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK8-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK8-NEXT: br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK8: omp_offload.failed: // CHECK8: omp_offload.failed:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l195(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK8: omp_offload.cont: // CHECK8: omp_offload.cont:
// CHECK8-NEXT: br label [[OMP_IF_END:%.*]] // CHECK8-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK8: omp_if.else: // CHECK8: omp_if.else:
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4 // CHECK8-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP16]], align 4
// CHECK8-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK8-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK8-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK8-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK8-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK8-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK8-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK8-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK8-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK8-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.15, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.16, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0) // CHECK8-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 1, i32 0)
// CHECK8-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK8-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK8-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK8-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK8: omp_offload.failed: // CHECK8: omp_offload.failed:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l178(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK8: omp_offload.cont: // CHECK8: omp_offload.cont:
// CHECK8-NEXT: br label [[OMP_IF_END:%.*]] // CHECK8-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK8: omp_if.else: // CHECK8: omp_if.else:
Show All 37 Lines
// CHECK8-NEXT: [[CONV3:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__CASTED]] to i8* // CHECK8-NEXT: [[CONV3:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__CASTED]] to i8*
// CHECK8-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 // CHECK8-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
// CHECK8-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1 // CHECK8-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1
// CHECK8-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK8-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK8-NEXT: [[TMP9:%.*]] = load i8, i8* [[CONV]], align 1 // CHECK8-NEXT: [[TMP9:%.*]] = load i8, i8* [[CONV]], align 1
// CHECK8-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP9]] to i1 // CHECK8-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP9]] to i1
// CHECK8-NEXT: br i1 [[TOBOOL4]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK8-NEXT: br i1 [[TOBOOL4]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
// CHECK8: omp_if.then: // CHECK8: omp_if.then:
// CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*, i32)* @.omp_outlined..9 to void (i32*, i32*, ...)*), %struct.S1* [[TMP1]], i32 [[TMP6]], i32 [[TMP2]], i32 [[TMP3]], i16* [[TMP4]], i32 [[TMP8]]) // CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*, i32)* @.omp_outlined..10 to void (i32*, i32*, ...)*), %struct.S1* [[TMP1]], i32 [[TMP6]], i32 [[TMP2]], i32 [[TMP3]], i16* [[TMP4]], i32 [[TMP8]])
// CHECK8-NEXT: br label [[OMP_IF_END:%.*]] // CHECK8-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK8: omp_if.else: // CHECK8: omp_if.else:
// CHECK8-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]]) // CHECK8-NEXT: call void @__kmpc_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]])
// CHECK8-NEXT: store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4 // CHECK8-NEXT: store i32 [[TMP0]], i32* [[DOTTHREADID_TEMP_]], align 4
// CHECK8-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4 // CHECK8-NEXT: store i32 0, i32* [[DOTBOUND_ZERO_ADDR]], align 4
// CHECK8-NEXT: call void @.omp_outlined..9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]], %struct.S1* [[TMP1]], i32 [[TMP6]], i32 [[TMP2]], i32 [[TMP3]], i16* [[TMP4]], i32 [[TMP8]]) #[[ATTR4]] // CHECK8-NEXT: call void @.omp_outlined..10(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTBOUND_ZERO_ADDR]], %struct.S1* [[TMP1]], i32 [[TMP6]], i32 [[TMP2]], i32 [[TMP3]], i16* [[TMP4]], i32 [[TMP8]]) #[[ATTR4]]
// CHECK8-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]]) // CHECK8-NEXT: call void @__kmpc_end_serialized_parallel(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]])
// CHECK8-NEXT: br label [[OMP_IF_END]] // CHECK8-NEXT: br label [[OMP_IF_END]]
// CHECK8: omp_if.end: // CHECK8: omp_if.end:
// CHECK8-NEXT: ret void // CHECK8-NEXT: ret void
// //
// //
// CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK8-NEXT: entry: // CHECK8-NEXT: entry:
// CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK8-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK8-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 169 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK8-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK8-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK8-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK8-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2 // CHECK8-NEXT: store i16 [[TMP3]], i16* [[CONV2]], align 2
// CHECK8-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK8-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK8-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK8-NEXT: [[TMP5:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK8-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK8-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK8-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1 // CHECK8-NEXT: store i8 [[TMP5]], i8* [[CONV3]], align 1
// CHECK8-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK8-NEXT: [[TMP6:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]]) // CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [10 x i32]* [[TMP0]])
// CHECK8-NEXT: ret void // CHECK8-NEXT: ret void
// //
// //
// CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK8-NEXT: entry: // CHECK8-NEXT: entry:
// CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK8-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
Show All 26 Lines
// CHECK8-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK8-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK8-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK8-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK8-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK8-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK8-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK8-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK8-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK8-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK8-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK8-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK8-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK8-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK8-NEXT: ret void // CHECK8-NEXT: ret void
// //
// //
// CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..14 // CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..16
// CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK8-NEXT: entry: // CHECK8-NEXT: entry:
// CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK8-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK8-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK8-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 14,992 Lines • Show Last 20 Lines

clang/test/OpenMP/target_simd_codegen.cpp

Show First 20 Lines • Show All 77 Lines • ▼ Show 20 Lines
// We have 8 target regions, but only 7 that actually will generate offloading // We have 8 target regions, but only 7 that actually will generate offloading
// code, only 6 will have mapped arguments, and only 4 have all-constant map // code, only 6 will have mapped arguments, and only 4 have all-constant map
// sizes. // sizes.
// CHECK-DAG: [[SIZET2:@.+]] = private unnamed_addr constant [3 x i64] [i64 2, i64 4, i64 4] // CHECK-DAG: [[SIZET2:@.+]] = private unnamed_addr constant [3 x i64] [i64 2, i64 4, i64 4]
// CHECK-DAG: [[MAPT2:@.+]] = private unnamed_addr constant [3 x i64] [i64 800, i64 800, i64 800] // CHECK-DAG: [[MAPT2:@.+]] = private unnamed_addr constant [3 x i64] [i64 800, i64 800, i64 800]
// CHECK-DAG: [[SIZET3:@.+]] = private unnamed_addr constant [2 x i64] [i64 4, i64 2] // CHECK-DAG: [[SIZET3:@.+]] = private unnamed_addr constant [2 x i64] [i64 4, i64 2]
// CHECK-DAG: [[MAPT3:@.+]] = private unnamed_addr constant [2 x i64] [i64 800, i64 800] // CHECK-DAG: [[MAPT3:@.+]] = private unnamed_addr constant [2 x i64] [i64 800, i64 800]
// CHECK-DAG: [[SIZET4:@.+]] = private unnamed_addr global [9 x i64] [i64 4, i64 40, i64 {{8|4}}, i64 0, i64 400, i64 {{8|4}}, i64 {{8|4}}, i64 0, i64 {{16|12}}]
// CHECK-DAG: [[MAPT4:@.+]] = private unnamed_addr constant [9 x i64] [i64 800, i64 547, i64 800, i64 547, i64 547, i64 800, i64 800, i64 547, i64 547] // CHECK-DAG: [[MAPT4:@.+]] = private unnamed_addr constant [9 x i64] [i64 800, i64 547, i64 800, i64 547, i64 547, i64 800, i64 800, i64 547, i64 547]
// CHECK-DAG: [[SIZET5:@.+]] = private unnamed_addr constant [3 x i64] [i64 4, i64 2, i64 40] // CHECK-DAG: [[SIZET5:@.+]] = private unnamed_addr constant [3 x i64] [i64 4, i64 2, i64 40]
// CHECK-DAG: [[MAPT5:@.+]] = private unnamed_addr constant [3 x i64] [i64 800, i64 800, i64 547] // CHECK-DAG: [[MAPT5:@.+]] = private unnamed_addr constant [3 x i64] [i64 800, i64 800, i64 547]
// CHECK-DAG: [[SIZET6:@.+]] = private unnamed_addr constant [4 x i64] [i64 4, i64 2, i64 1, i64 40] // CHECK-DAG: [[SIZET6:@.+]] = private unnamed_addr constant [4 x i64] [i64 4, i64 2, i64 1, i64 40]
// CHECK-DAG: [[MAPT6:@.+]] = private unnamed_addr constant [4 x i64] [i64 800, i64 800, i64 800, i64 547] // CHECK-DAG: [[MAPT6:@.+]] = private unnamed_addr constant [4 x i64] [i64 800, i64 800, i64 800, i64 547]
// OMP45-DAG: [[SIZET7:@.+]] = private unnamed_addr global [5 x i64] [i64 {{8|4}}, i64 4, i64 {{8|4}}, i64 {{8|4}}, i64 0]
// OMP50-DAG: [[SIZET7:@.+]] = private unnamed_addr global [6 x i64] [i64 {{8|4}}, i64 4, i64 {{8|4}}, i64 {{8|4}}, i64 0, i64 1]
// OMP45-DAG: [[MAPT7:@.+]] = private unnamed_addr constant [5 x i64] [i64 547, i64 800, i64 800, i64 800, i64 547] // OMP45-DAG: [[MAPT7:@.+]] = private unnamed_addr constant [5 x i64] [i64 547, i64 800, i64 800, i64 800, i64 547]
// OMP50-DAG: [[MAPT7:@.+]] = private unnamed_addr constant [6 x i64] [i64 547, i64 800, i64 800, i64 800, i64 547, i64 800] // OMP50-DAG: [[MAPT7:@.+]] = private unnamed_addr constant [6 x i64] [i64 547, i64 800, i64 800, i64 800, i64 547, i64 800]
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
// CHECK-DAG: @{{.*}} = weak constant i8 0 // CHECK-DAG: @{{.*}} = weak constant i8 0
▲ Show 20 Lines • Show All 133 Lines • ▼ Show 20 Linesint foo(int n) {
// CHECK-64: [[BNSIZE:%.+]] = mul nuw i64 [[VLA0:%.+]], 4 // CHECK-64: [[BNSIZE:%.+]] = mul nuw i64 [[VLA0:%.+]], 4
// CHECK-32: [[BNSZSIZE:%.+]] = mul nuw i32 [[VLA0:%.+]], 4 // CHECK-32: [[BNSZSIZE:%.+]] = mul nuw i32 [[VLA0:%.+]], 4
// CHECK-32: [[BNSIZE:%.+]] = sext i32 [[BNSZSIZE]] to i64 // CHECK-32: [[BNSIZE:%.+]] = sext i32 [[BNSZSIZE]] to i64
// CHECK: [[CNELEMSIZE2:%.+]] = mul nuw i[[SZ]] 5, [[VLA1:%.+]] // CHECK: [[CNELEMSIZE2:%.+]] = mul nuw i[[SZ]] 5, [[VLA1:%.+]]
// CHECK-64: [[CNSIZE:%.+]] = mul nuw i64 [[CNELEMSIZE2]], 8 // CHECK-64: [[CNSIZE:%.+]] = mul nuw i64 [[CNELEMSIZE2]], 8
// CHECK-32: [[CNSZSIZE:%.+]] = mul nuw i32 [[CNELEMSIZE2]], 8 // CHECK-32: [[CNSZSIZE:%.+]] = mul nuw i32 [[CNELEMSIZE2]], 8
// CHECK-32: [[CNSIZE:%.+]] = sext i32 [[CNSZSIZE]] to i64 // CHECK-32: [[CNSIZE:%.+]] = sext i32 [[CNSZSIZE]] to i64
// CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 9, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* [[SR:%[^,]+]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[MAPT4]], i32 0, i32 0), i8** null, i8** null, i32 1, i32 1) // CHECK-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 9, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[SIZET4]], i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[MAPT4]], i32 0, i32 0), i8** null, i8** null, i32 1, i32 1)
// CHECK-DAG: [[BPR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP:%[^,]+]], i32 0, i32 0 // CHECK-DAG: [[BPR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP:%[^,]+]], i32 0, i32 0
// CHECK-DAG: [[PR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P:%[^,]+]], i32 0, i32 0 // CHECK-DAG: [[PR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P:%[^,]+]], i32 0, i32 0
// CHECK-DAG: [[SR]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S:%[^,]+]], i32 0, i32 0
// CHECK-DAG: [[SADDR0:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX0:[0-9]+]] // CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX0:[0-9]+]]
// CHECK-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX0]]
// CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX0]] // CHECK-DAG: [[PADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX0]]
// CHECK-DAG: [[SADDR1:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX1:[0-9]+]] // CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX1:[0-9]+]]
// CHECK-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX1]]
// CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX1]] // CHECK-DAG: [[PADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX1]]
// CHECK-DAG: [[SADDR2:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX2:[0-9]+]] // CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX2:[0-9]+]]
// CHECK-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX2]]
// CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX2]] // CHECK-DAG: [[PADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX2]]
// CHECK-DAG: [[SADDR3:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX3:[0-9]+]] // CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX3:[0-9]+]]
// CHECK-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX3]]
// CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX3]] // CHECK-DAG: [[PADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX3]]
// CHECK-DAG: [[SADDR4:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX4:[0-9]+]] // CHECK-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX4:[0-9]+]]
// CHECK-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX4]]
// CHECK-DAG: [[PADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX4]] // CHECK-DAG: [[PADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX4]]
// CHECK-DAG: [[SADDR5:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX5:[0-9]+]] // CHECK-DAG: [[BPADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX5:[0-9]+]]
// CHECK-DAG: [[BPADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX5]]
// CHECK-DAG: [[PADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX5]] // CHECK-DAG: [[PADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX5]]
// CHECK-DAG: [[SADDR6:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX6:[0-9]+]] // CHECK-DAG: [[BPADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX6:[0-9]+]]
// CHECK-DAG: [[BPADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX6]]
// CHECK-DAG: [[PADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX6]] // CHECK-DAG: [[PADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX6]]
// CHECK-DAG: [[SADDR7:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX7:[0-9]+]] // CHECK-DAG: [[BPADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX7:[0-9]+]]
// CHECK-DAG: [[BPADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX7]]
// CHECK-DAG: [[PADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX7]] // CHECK-DAG: [[PADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX7]]
// CHECK-DAG: [[SADDR8:%.+]] = getelementptr inbounds [9 x i64], [9 x i64]* [[S]], i32 0, i32 [[IDX8:[0-9]+]] // CHECK-DAG: [[BPADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX8:[0-9]+]]
// CHECK-DAG: [[BPADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX8]]
// CHECK-DAG: [[PADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX8]] // CHECK-DAG: [[PADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX8]]
// The names below are not necessarily consistent with the names used for the // The names below are not necessarily consistent with the names used for the
// addresses above as some are repeated. // addresses above as some are repeated.
// CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CBPADDR0:%.+]], // CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CBPADDR0:%.+]],
// CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CPADDR0:%.+]], // CHECK-DAG: store i[[SZ]] [[A_CVAL]], i[[SZ]]* [[CPADDR0:%.+]],
// CHECK-DAG: [[CBPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CBPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: [[CPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CPADDR0]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: store i64 4, i64* {{%[^,]+}}
// CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CBPADDR1:%.+]], // CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CBPADDR1:%.+]],
// CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CPADDR1:%.+]], // CHECK-DAG: store [10 x float]* %{{.+}}, [10 x float]** [[CPADDR1:%.+]],
// CHECK-DAG: [[CBPADDR1]] = bitcast i8** {{%[^,]+}} to [10 x float]** // CHECK-DAG: [[CBPADDR1]] = bitcast i8** {{%[^,]+}} to [10 x float]**
// CHECK-DAG: [[CPADDR1]] = bitcast i8** {{%[^,]+}} to [10 x float]** // CHECK-DAG: [[CPADDR1]] = bitcast i8** {{%[^,]+}} to [10 x float]**
// CHECK-DAG: store i64 40, i64* {{%[^,]+}}
// CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR2:%.+]], // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR2:%.+]],
// CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR2:%.+]], // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR2:%.+]],
// CHECK-DAG: [[CBPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CBPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: [[CPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}}
// CHECK-DAG: store float* %{{.+}}, float** [[CBPADDR3:%.+]], // CHECK-DAG: store float* %{{.+}}, float** [[CBPADDR3:%.+]],
// CHECK-DAG: store float* %{{.+}}, float** [[CPADDR3:%.+]], // CHECK-DAG: store float* %{{.+}}, float** [[CPADDR3:%.+]],
// CHECK-DAG: [[CBPADDR3]] = bitcast i8** {{%[^,]+}} to float** // CHECK-DAG: [[CBPADDR3]] = bitcast i8** {{%[^,]+}} to float**
// CHECK-DAG: [[CPADDR3]] = bitcast i8** {{%[^,]+}} to float** // CHECK-DAG: [[CPADDR3]] = bitcast i8** {{%[^,]+}} to float**
// CHECK-DAG: store i64 [[BNSIZE]], i64* {{%[^,]+}} // CHECK-DAG: store i64 [[BNSIZE]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[SIZET4]], i32 0, i32 3),
// CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CBPADDR4:%.+]], // CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CBPADDR4:%.+]],
// CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CPADDR4:%.+]], // CHECK-DAG: store [5 x [10 x double]]* %{{.+}}, [5 x [10 x double]]** [[CPADDR4:%.+]],
// CHECK-DAG: [[CBPADDR4]] = bitcast i8** {{%[^,]+}} to [5 x [10 x double]]** // CHECK-DAG: [[CBPADDR4]] = bitcast i8** {{%[^,]+}} to [5 x [10 x double]]**
// CHECK-DAG: [[CPADDR4]] = bitcast i8** {{%[^,]+}} to [5 x [10 x double]]** // CHECK-DAG: [[CPADDR4]] = bitcast i8** {{%[^,]+}} to [5 x [10 x double]]**
// CHECK-DAG: store i64 400, i64* {{%[^,]+}}
// CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CBPADDR5:%.+]], // CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CBPADDR5:%.+]],
// CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CPADDR5:%.+]], // CHECK-DAG: store i[[SZ]] 5, i[[SZ]]* [[CPADDR5:%.+]],
// CHECK-DAG: [[CBPADDR5]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CBPADDR5]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: [[CPADDR5]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CPADDR5]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}}
// CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR6:%.+]], // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR6:%.+]],
// CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR6:%.+]], // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR6:%.+]],
// CHECK-DAG: [[CBPADDR6]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CBPADDR6]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: [[CPADDR6]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CPADDR6]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}}
// CHECK-DAG: store double* %{{.+}}, double** [[CBPADDR7:%.+]], // CHECK-DAG: store double* %{{.+}}, double** [[CBPADDR7:%.+]],
// CHECK-DAG: store double* %{{.+}}, double** [[CPADDR7:%.+]], // CHECK-DAG: store double* %{{.+}}, double** [[CPADDR7:%.+]],
// CHECK-DAG: [[CBPADDR7]] = bitcast i8** {{%[^,]+}} to double** // CHECK-DAG: [[CBPADDR7]] = bitcast i8** {{%[^,]+}} to double**
// CHECK-DAG: [[CPADDR7]] = bitcast i8** {{%[^,]+}} to double** // CHECK-DAG: [[CPADDR7]] = bitcast i8** {{%[^,]+}} to double**
// CHECK-DAG: store i64 [[CNSIZE]], i64* {{%[^,]+}} // CHECK-DAG: store i64 [[CNSIZE]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[SIZET4]], i32 0, i32 7),
// CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CBPADDR8:%.+]], // CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CBPADDR8:%.+]],
// CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CPADDR8:%.+]], // CHECK-DAG: store [[TT]]* %{{.+}}, [[TT]]** [[CPADDR8:%.+]],
// CHECK-DAG: [[CBPADDR8]] = bitcast i8** {{%[^,]+}} to [[TT]]** // CHECK-DAG: [[CBPADDR8]] = bitcast i8** {{%[^,]+}} to [[TT]]**
// CHECK-DAG: [[CPADDR8]] = bitcast i8** {{%[^,]+}} to [[TT]]** // CHECK-DAG: [[CPADDR8]] = bitcast i8** {{%[^,]+}} to [[TT]]**
// CHECK-DAG: store i64 {{12|16}}, i64* {{%[^,]+}}
// CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 // CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0
// CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]] // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]]
// CHECK: [[FAIL]] // CHECK: [[FAIL]]
// CHECK: call void [[HVT4:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) // CHECK: call void [[HVT4:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}})
// CHECK-NEXT: br label %[[END]] // CHECK-NEXT: br label %[[END]]
// CHECK: [[END]] // CHECK: [[END]]
▲ Show 20 Lines • Show All 191 Lines • ▼ Show 20 Lines
// CHECK: br i1 [[IF]], label %[[TRY:[^,]+]], label %[[FAIL:[^,]+]] // CHECK: br i1 [[IF]], label %[[TRY:[^,]+]], label %[[FAIL:[^,]+]]
// CHECK: [[TRY]] // CHECK: [[TRY]]
// We capture 2 VLA sizes in this target region // We capture 2 VLA sizes in this target region
// CHECK: [[CELEMSIZE2:%.+]] = mul nuw i[[SZ]] 2, [[VLA0:%.+]] // CHECK: [[CELEMSIZE2:%.+]] = mul nuw i[[SZ]] 2, [[VLA0:%.+]]
// CHECK-64: [[CSIZE:%.+]] = mul nuw i64 [[CELEMSIZE2]], 2 // CHECK-64: [[CSIZE:%.+]] = mul nuw i64 [[CELEMSIZE2]], 2
// CHECK-32: [[CSZSIZE:%.+]] = mul nuw i32 [[CELEMSIZE2]], 2 // CHECK-32: [[CSZSIZE:%.+]] = mul nuw i32 [[CELEMSIZE2]], 2
// CHECK-32: [[CSIZE:%.+]] = sext i32 [[CSZSIZE]] to i64 // CHECK-32: [[CSIZE:%.+]] = sext i32 [[CSZSIZE]] to i64
// OMP45-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 5, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* [[SR:%[^,]+]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPT7]], i32 0, i32 0), i8** null, i8** null, i32 1, i32 1) // OMP45-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 5, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[SIZET7]], i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPT7]], i32 0, i32 0), i8** null, i8** null, i32 1, i32 1)
// OMP50-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 6, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* [[SR:%[^,]+]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* [[MAPT7]], i32 0, i32 0), i8** null, i8** null, i32 1, i32 1) // OMP50-DAG: [[RET:%.+]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 6, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* [[SIZET7]], i32 0, i32 0), i64* getelementptr inbounds ([6 x i64], [6 x i64]* [[MAPT7]], i32 0, i32 0), i8** null, i8** null, i32 1, i32 1)
// OMP45-DAG: [[BPR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP:%.+]], i32 0, i32 0 // OMP45-DAG: [[BPR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP:%.+]], i32 0, i32 0
// OMP45-DAG: [[PR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P:%.+]], i32 0, i32 0 // OMP45-DAG: [[PR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P:%.+]], i32 0, i32 0
// OMP45-DAG: [[SR]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S:%.+]], i32 0, i32 0 // OMP45-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX0:[0-9]+]]
// OMP45-DAG: [[SADDR0:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S]], i32 [[IDX0:[0-9]+]]
// OMP45-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX0]]
// OMP45-DAG: [[PADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX0]] // OMP45-DAG: [[PADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX0]]
// OMP45-DAG: [[SADDR1:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S]], i32 [[IDX1:[0-9]+]] // OMP45-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX1:[0-9]+]]
// OMP45-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX1]]
// OMP45-DAG: [[PADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX1]] // OMP45-DAG: [[PADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX1]]
// OMP45-DAG: [[SADDR2:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S]], i32 [[IDX2:[0-9]+]] // OMP45-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX2:[0-9]+]]
// OMP45-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX2]]
// OMP45-DAG: [[PADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX2]] // OMP45-DAG: [[PADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX2]]
// OMP45-DAG: [[SADDR3:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S]], i32 [[IDX3:[0-9]+]] // OMP45-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX3:[0-9]+]]
// OMP45-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX3]]
// OMP45-DAG: [[PADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX3]] // OMP45-DAG: [[PADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX3]]
// OMP45-DAG: [[SADDR4:%.+]] = getelementptr inbounds [5 x i64], [5 x i64]* [[S]], i32 [[IDX4:[0-9]+]] // OMP45-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX4:[0-9]+]]
// OMP45-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX4]]
// OMP45-DAG: [[PADDR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX4]] // OMP45-DAG: [[PADDR4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX4]]
// OMP50-DAG: [[BPR]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP:%.+]], i32 0, i32 0 // OMP50-DAG: [[BPR]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP:%.+]], i32 0, i32 0
// OMP50-DAG: [[PR]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P:%.+]], i32 0, i32 0 // OMP50-DAG: [[PR]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P:%.+]], i32 0, i32 0
// OMP50-DAG: [[SR]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S:%.+]], i32 0, i32 0 // OMP50-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX0:[0-9]+]]
// OMP50-DAG: [[SADDR0:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX0:[0-9]+]]
// OMP50-DAG: [[BPADDR0:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX0]]
// OMP50-DAG: [[PADDR0:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX0]] // OMP50-DAG: [[PADDR0:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX0]]
// OMP50-DAG: [[SADDR1:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX1:[0-9]+]] // OMP50-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX1:[0-9]+]]
// OMP50-DAG: [[BPADDR1:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX1]]
// OMP50-DAG: [[PADDR1:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX1]] // OMP50-DAG: [[PADDR1:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX1]]
// OMP50-DAG: [[SADDR2:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX2:[0-9]+]] // OMP50-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX2:[0-9]+]]
// OMP50-DAG: [[BPADDR2:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX2]]
// OMP50-DAG: [[PADDR2:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX2]] // OMP50-DAG: [[PADDR2:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX2]]
// OMP50-DAG: [[SADDR3:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX3:[0-9]+]] // OMP50-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX3:[0-9]+]]
// OMP50-DAG: [[BPADDR3:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX3]]
// OMP50-DAG: [[PADDR3:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX3]] // OMP50-DAG: [[PADDR3:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX3]]
// OMP50-DAG: [[SADDR4:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX4:[0-9]+]] // OMP50-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX4:[0-9]+]]
// OMP50-DAG: [[BPADDR4:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX4]]
// OMP50-DAG: [[PADDR4:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX4]] // OMP50-DAG: [[PADDR4:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX4]]
// OMP50-DAG: [[SADDR5:%.+]] = getelementptr inbounds [6 x i64], [6 x i64]* [[S]], i32 [[IDX5:[0-9]+]] // OMP50-DAG: [[BPADDR5:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX5:[0-9]+]]
// OMP50-DAG: [[BPADDR5:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[BP]], i32 [[IDX5]]
// OMP50-DAG: [[PADDR5:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX5]] // OMP50-DAG: [[PADDR5:%.+]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[P]], i32 [[IDX5]]
// The names below are not necessarily consistent with the names used for the // The names below are not necessarily consistent with the names used for the
// addresses above as some are repeated. // addresses above as some are repeated.
// CHECK-DAG: store [[S1]]* %{{.+}}, [[S1]]** [[CBPADDR0:%.+]], // CHECK-DAG: store [[S1]]* %{{.+}}, [[S1]]** [[CBPADDR0:%.+]],
// CHECK-DAG: store [[S1]]* %{{.+}}, [[S1]]** [[CPADDR0:%.+]], // CHECK-DAG: store [[S1]]* %{{.+}}, [[S1]]** [[CPADDR0:%.+]],
// CHECK-DAG: [[CBPADDR0]] = bitcast i8** {{%[^,]+}} to [[S1]]** // CHECK-DAG: [[CBPADDR0]] = bitcast i8** {{%[^,]+}} to [[S1]]**
// CHECK-DAG: [[CPADDR0]] = bitcast i8** {{%[^,]+}} to [[S1]]** // CHECK-DAG: [[CPADDR0]] = bitcast i8** {{%[^,]+}} to [[S1]]**
// CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}}
// CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CBPADDR1:%.+]], // CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CBPADDR1:%.+]],
// CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CPADDR1:%.+]], // CHECK-DAG: store i[[SZ]] [[B_CVAL]], i[[SZ]]* [[CPADDR1:%.+]],
// CHECK-DAG: [[CBPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CBPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: [[CPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CPADDR1]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: store i64 4, i64* {{%[^,]+}}
// CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CBPADDR2:%.+]], // CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CBPADDR2:%.+]],
// CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CPADDR2:%.+]], // CHECK-DAG: store i[[SZ]] 2, i[[SZ]]* [[CPADDR2:%.+]],
// CHECK-DAG: [[CBPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CBPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: [[CPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CPADDR2]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}}
// CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR3:%.+]], // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CBPADDR3:%.+]],
// CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR3:%.+]], // CHECK-DAG: store i[[SZ]] %{{.+}}, i[[SZ]]* [[CPADDR3:%.+]],
// CHECK-DAG: [[CBPADDR3]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CBPADDR3]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: [[CPADDR3]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // CHECK-DAG: [[CPADDR3]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// CHECK-DAG: store i64 {{4|8}}, i64* {{%[^,]+}}
// CHECK-DAG: store i16* %{{.+}}, i16** [[CBPADDR4:%.+]], // CHECK-DAG: store i16* %{{.+}}, i16** [[CBPADDR4:%.+]],
// CHECK-DAG: store i16* %{{.+}}, i16** [[CPADDR4:%.+]], // CHECK-DAG: store i16* %{{.+}}, i16** [[CPADDR4:%.+]],
// CHECK-DAG: [[CBPADDR4]] = bitcast i8** {{%[^,]+}} to i16** // CHECK-DAG: [[CBPADDR4]] = bitcast i8** {{%[^,]+}} to i16**
// CHECK-DAG: [[CPADDR4]] = bitcast i8** {{%[^,]+}} to i16** // CHECK-DAG: [[CPADDR4]] = bitcast i8** {{%[^,]+}} to i16**
// CHECK-DAG: store i64 [[CSIZE]], i64* {{%[^,]+}} // CHECK-DAG: store i64 [[CSIZE]], i64* getelementptr inbounds ([{{.+}} x i64], [{{.+}} x i64]* [[SIZET7]], i32 0, i32 4),
// OMP50-DAG: store i[[SZ]] [[SIMD_COND]], i[[SZ]]* [[CBPADDR5:%.+]], // OMP50-DAG: store i[[SZ]] [[SIMD_COND]], i[[SZ]]* [[CBPADDR5:%.+]],
// OMP50-DAG: store i[[SZ]] [[SIMD_COND]], i[[SZ]]* [[CPADDR5:%.+]], // OMP50-DAG: store i[[SZ]] [[SIMD_COND]], i[[SZ]]* [[CPADDR5:%.+]],
// OMP50-DAG: [[CBPADDR5]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // OMP50-DAG: [[CBPADDR5]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// OMP50-DAG: [[CPADDR5]] = bitcast i8** {{%[^,]+}} to i[[SZ]]* // OMP50-DAG: [[CPADDR5]] = bitcast i8** {{%[^,]+}} to i[[SZ]]*
// OMP50-DAG: store i64 1, i64* {{%[^,]+}}
// CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0 // CHECK-NEXT: [[ERROR:%.+]] = icmp ne i32 [[RET]], 0
// CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]] // CHECK-NEXT: br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]]
// CHECK: [[FAIL]] // CHECK: [[FAIL]]
// OMP45: call void [[HVT7:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) // OMP45: call void [[HVT7:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}})
// OMP50: call void [[HVT7:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}) // OMP50: call void [[HVT7:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}})
// CHECK-NEXT: br label %[[END]] // CHECK-NEXT: br label %[[END]]
▲ Show 20 Lines • Show All 160 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_codegen.cpp

This file has a very large number of changes (2,968 lines). Show File Contents

clang/test/OpenMP/target_teams_distribute_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 327 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[_TMP19:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP19:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A_CASTED23:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_CASTED23:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED25:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED25:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [10 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [10 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [10 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [10 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [10 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [10 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK1-NEXT: [[_TMP32:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP32:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 0, i32* [[A]], align 4 // CHECK1-NEXT: store i32 0, i32* [[A]], align 4
// CHECK1-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK1-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK1-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
▲ Show 20 Lines • Show All 168 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP96:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK1-NEXT: [[TMP96:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK1-NEXT: [[TMP97:%.*]] = mul nuw i64 [[TMP96]], 8 // CHECK1-NEXT: [[TMP97:%.*]] = mul nuw i64 [[TMP96]], 8
// CHECK1-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0 // CHECK1-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK1-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64* // CHECK1-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64*
// CHECK1-NEXT: store i64 [[TMP91]], i64* [[TMP99]], align 8 // CHECK1-NEXT: store i64 [[TMP91]], i64* [[TMP99]], align 8
// CHECK1-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0 // CHECK1-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK1-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64* // CHECK1-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64*
// CHECK1-NEXT: store i64 [[TMP91]], i64* [[TMP101]], align 8 // CHECK1-NEXT: store i64 [[TMP91]], i64* [[TMP101]], align 8
// CHECK1-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK1-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 0
// CHECK1-NEXT: store i64 4, i64* [[TMP102]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK1-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 0 // CHECK1-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1
// CHECK1-NEXT: store i8* null, i8** [[TMP103]], align 8 // CHECK1-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]**
// CHECK1-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1 // CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 8
// CHECK1-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to [10 x float]** // CHECK1-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1
// CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP105]], align 8 // CHECK1-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to [10 x float]**
// CHECK1-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1 // CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP106]], align 8
// CHECK1-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to [10 x float]** // CHECK1-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 1
// CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP107]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP107]], align 8
// CHECK1-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK1-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2
// CHECK1-NEXT: store i64 40, i64* [[TMP108]], align 8 // CHECK1-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to i64*
// CHECK1-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 1 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP109]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP109]], align 8 // CHECK1-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2
// CHECK1-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2
// CHECK1-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i64* // CHECK1-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i64*
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP111]], align 8 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP111]], align 8
// CHECK1-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2 // CHECK1-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 2
// CHECK1-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i64* // CHECK1-NEXT: store i8* null, i8** [[TMP112]], align 8
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP113]], align 8 // CHECK1-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3
// CHECK1-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK1-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float**
// CHECK1-NEXT: store i64 8, i64* [[TMP114]], align 8 // CHECK1-NEXT: store float* [[VLA]], float** [[TMP114]], align 8
// CHECK1-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 2 // CHECK1-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3
// CHECK1-NEXT: store i8* null, i8** [[TMP115]], align 8 // CHECK1-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float**
// CHECK1-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3 // CHECK1-NEXT: store float* [[VLA]], float** [[TMP116]], align 8
// CHECK1-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to float** // CHECK1-NEXT: store i64 [[TMP95]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 3), align 8
// CHECK1-NEXT: store float* [[VLA]], float** [[TMP117]], align 8 // CHECK1-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 3
// CHECK1-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3 // CHECK1-NEXT: store i8* null, i8** [[TMP117]], align 8
// CHECK1-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to float** // CHECK1-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4
// CHECK1-NEXT: store float* [[VLA]], float** [[TMP119]], align 8 // CHECK1-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to [5 x [10 x double]]**
// CHECK1-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP119]], align 8
// CHECK1-NEXT: store i64 [[TMP95]], i64* [[TMP120]], align 8 // CHECK1-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4
// CHECK1-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 3 // CHECK1-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to [5 x [10 x double]]**
// CHECK1-NEXT: store i8* null, i8** [[TMP121]], align 8 // CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP121]], align 8
// CHECK1-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4 // CHECK1-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 4
// CHECK1-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to [5 x [10 x double]]** // CHECK1-NEXT: store i8* null, i8** [[TMP122]], align 8
// CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP123]], align 8 // CHECK1-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5
// CHECK1-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4 // CHECK1-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i64*
// CHECK1-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to [5 x [10 x double]]** // CHECK1-NEXT: store i64 5, i64* [[TMP124]], align 8
// CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP125]], align 8 // CHECK1-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5
// CHECK1-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK1-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64*
// CHECK1-NEXT: store i64 400, i64* [[TMP126]], align 8 // CHECK1-NEXT: store i64 5, i64* [[TMP126]], align 8
// CHECK1-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 4 // CHECK1-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 5
// CHECK1-NEXT: store i8* null, i8** [[TMP127]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP127]], align 8
// CHECK1-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5 // CHECK1-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6
// CHECK1-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i64* // CHECK1-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i64*
// CHECK1-NEXT: store i64 5, i64* [[TMP129]], align 8 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP129]], align 8
// CHECK1-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5 // CHECK1-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6
// CHECK1-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i64* // CHECK1-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i64*
// CHECK1-NEXT: store i64 5, i64* [[TMP131]], align 8 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP131]], align 8
// CHECK1-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK1-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 6
// CHECK1-NEXT: store i64 8, i64* [[TMP132]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP132]], align 8
// CHECK1-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 5 // CHECK1-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7
// CHECK1-NEXT: store i8* null, i8** [[TMP133]], align 8 // CHECK1-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double**
// CHECK1-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6 // CHECK1-NEXT: store double* [[VLA1]], double** [[TMP134]], align 8
// CHECK1-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i64* // CHECK1-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP135]], align 8 // CHECK1-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to double**
// CHECK1-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6 // CHECK1-NEXT: store double* [[VLA1]], double** [[TMP136]], align 8
// CHECK1-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i64* // CHECK1-NEXT: store i64 [[TMP97]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 7), align 8
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP137]], align 8 // CHECK1-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 7
// CHECK1-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK1-NEXT: store i8* null, i8** [[TMP137]], align 8
// CHECK1-NEXT: store i64 8, i64* [[TMP138]], align 8 // CHECK1-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8
// CHECK1-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 6 // CHECK1-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to %struct.TT**
// CHECK1-NEXT: store i8* null, i8** [[TMP139]], align 8 // CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP139]], align 8
// CHECK1-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7 // CHECK1-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8
// CHECK1-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to double** // CHECK1-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to %struct.TT**
// CHECK1-NEXT: store double* [[VLA1]], double** [[TMP141]], align 8 // CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP141]], align 8
// CHECK1-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7 // CHECK1-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 8
// CHECK1-NEXT: [[TMP143:%.*]] = bitcast i8** [[TMP142]] to double** // CHECK1-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK1-NEXT: store double* [[VLA1]], double** [[TMP143]], align 8 // CHECK1-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9
// CHECK1-NEXT: [[TMP144:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK1-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i64*
// CHECK1-NEXT: store i64 [[TMP97]], i64* [[TMP144]], align 8 // CHECK1-NEXT: store i64 [[TMP93]], i64* [[TMP144]], align 8
// CHECK1-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 7 // CHECK1-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9
// CHECK1-NEXT: store i8* null, i8** [[TMP145]], align 8 // CHECK1-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i64*
// CHECK1-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8 // CHECK1-NEXT: store i64 [[TMP93]], i64* [[TMP146]], align 8
// CHECK1-NEXT: [[TMP147:%.*]] = bitcast i8** [[TMP146]] to %struct.TT** // CHECK1-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 9
// CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP147]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP147]], align 8
// CHECK1-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8 // CHECK1-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK1-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to %struct.TT** // CHECK1-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP149]], align 8
// CHECK1-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK1-NEXT: store i64 16, i64* [[TMP150]], align 8
// CHECK1-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 8
// CHECK1-NEXT: store i8* null, i8** [[TMP151]], align 8
// CHECK1-NEXT: [[TMP152:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9
// CHECK1-NEXT: [[TMP153:%.*]] = bitcast i8** [[TMP152]] to i64*
// CHECK1-NEXT: store i64 [[TMP93]], i64* [[TMP153]], align 8
// CHECK1-NEXT: [[TMP154:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9
// CHECK1-NEXT: [[TMP155:%.*]] = bitcast i8** [[TMP154]] to i64*
// CHECK1-NEXT: store i64 [[TMP93]], i64* [[TMP155]], align 8
// CHECK1-NEXT: [[TMP156:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK1-NEXT: store i64 4, i64* [[TMP156]], align 8
// CHECK1-NEXT: [[TMP157:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 9
// CHECK1-NEXT: store i8* null, i8** [[TMP157]], align 8
// CHECK1-NEXT: [[TMP158:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK1-NEXT: [[TMP159:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK1-NEXT: [[TMP160:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK1-NEXT: [[TMP161:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP158]], i8** [[TMP159]], i64* [[TMP160]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP148]], i8** [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK1-NEXT: [[TMP162:%.*]] = icmp ne i32 [[TMP161]], 0 // CHECK1-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK1-NEXT: br i1 [[TMP162]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]] // CHECK1-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK1: omp_offload.failed33: // CHECK1: omp_offload.failed33:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT34]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK1: omp_offload.cont34: // CHECK1: omp_offload.cont34:
// CHECK1-NEXT: br label [[OMP_IF_END36:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END36:%.*]]
// CHECK1: omp_if.else35: // CHECK1: omp_if.else35:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]]
// CHECK1-NEXT: br label [[OMP_IF_END36]] // CHECK1-NEXT: br label [[OMP_IF_END36]]
// CHECK1: omp_if.end36: // CHECK1: omp_if.end36:
// CHECK1-NEXT: [[TMP163:%.*]] = load i32, i32* [[A]], align 4 // CHECK1-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4
// CHECK1-NEXT: [[TMP164:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP164]]) // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP153]])
// CHECK1-NEXT: ret i32 [[TMP163]] // CHECK1-NEXT: ret i32 [[TMP152]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK1-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK1-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 678 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK1-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK1-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK1-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 15 Lines
// CHECK1-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK1-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK1-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK1-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK1-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK1-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK1-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK1-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK1-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK1-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK1-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK1-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK1-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK1-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK1-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK1-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK1-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK1-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK1-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK1-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 8
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK1-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK1-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK1-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK1-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK1-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK1-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]]
// CHECK1-NEXT: br label [[OMP_IF_END]] // CHECK1-NEXT: br label [[OMP_IF_END]]
// CHECK1: omp_if.end: // CHECK1: omp_if.end:
// CHECK1-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK1-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK1-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK1-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK1-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK1-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK1-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK1-NEXT: ret i32 [[ADD4]] // CHECK1-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK1-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK1-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1
// CHECK1-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK1-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK1-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1 // CHECK1-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1
// CHECK1-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK1-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK1-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK1-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK1-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1 // CHECK1-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1
// CHECK1-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64 // CHECK1-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK1-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK1-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK1-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK1-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
Show All 22 Lines
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2 // CHECK1-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2
// CHECK1-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK1-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2 // CHECK1-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2
// CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK1-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK1-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK1-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK1-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1 // CHECK1-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1
// CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 129 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[_TMP19:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[_TMP19:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A_CASTED23:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_CASTED23:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTCAPTURE_EXPR__CASTED25:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__CASTED25:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [10 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [10 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [10 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [10 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [10 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [10 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK2-NEXT: [[_TMP32:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[_TMP32:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 0, i32* [[A]], align 4 // CHECK2-NEXT: store i32 0, i32* [[A]], align 4
// CHECK2-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK2-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK2-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
▲ Show 20 Lines • Show All 168 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP96:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK2-NEXT: [[TMP96:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK2-NEXT: [[TMP97:%.*]] = mul nuw i64 [[TMP96]], 8 // CHECK2-NEXT: [[TMP97:%.*]] = mul nuw i64 [[TMP96]], 8
// CHECK2-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0 // CHECK2-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK2-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64* // CHECK2-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64*
// CHECK2-NEXT: store i64 [[TMP91]], i64* [[TMP99]], align 8 // CHECK2-NEXT: store i64 [[TMP91]], i64* [[TMP99]], align 8
// CHECK2-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0 // CHECK2-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK2-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64* // CHECK2-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64*
// CHECK2-NEXT: store i64 [[TMP91]], i64* [[TMP101]], align 8 // CHECK2-NEXT: store i64 [[TMP91]], i64* [[TMP101]], align 8
// CHECK2-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK2-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 0
// CHECK2-NEXT: store i64 4, i64* [[TMP102]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK2-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 0 // CHECK2-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP103]], align 8 // CHECK2-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]**
// CHECK2-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1 // CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 8
// CHECK2-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to [10 x float]** // CHECK2-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1
// CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP105]], align 8 // CHECK2-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to [10 x float]**
// CHECK2-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1 // CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP106]], align 8
// CHECK2-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to [10 x float]** // CHECK2-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 1
// CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP107]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP107]], align 8
// CHECK2-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK2-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2
// CHECK2-NEXT: store i64 40, i64* [[TMP108]], align 8 // CHECK2-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to i64*
// CHECK2-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 1 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP109]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP109]], align 8 // CHECK2-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2
// CHECK2-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2
// CHECK2-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i64* // CHECK2-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i64*
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP111]], align 8 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP111]], align 8
// CHECK2-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2 // CHECK2-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 2
// CHECK2-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i64* // CHECK2-NEXT: store i8* null, i8** [[TMP112]], align 8
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP113]], align 8 // CHECK2-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3
// CHECK2-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK2-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float**
// CHECK2-NEXT: store i64 8, i64* [[TMP114]], align 8 // CHECK2-NEXT: store float* [[VLA]], float** [[TMP114]], align 8
// CHECK2-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 2 // CHECK2-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3
// CHECK2-NEXT: store i8* null, i8** [[TMP115]], align 8 // CHECK2-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float**
// CHECK2-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3 // CHECK2-NEXT: store float* [[VLA]], float** [[TMP116]], align 8
// CHECK2-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to float** // CHECK2-NEXT: store i64 [[TMP95]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 3), align 8
// CHECK2-NEXT: store float* [[VLA]], float** [[TMP117]], align 8 // CHECK2-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 3
// CHECK2-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3 // CHECK2-NEXT: store i8* null, i8** [[TMP117]], align 8
// CHECK2-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to float** // CHECK2-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4
// CHECK2-NEXT: store float* [[VLA]], float** [[TMP119]], align 8 // CHECK2-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to [5 x [10 x double]]**
// CHECK2-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP119]], align 8
// CHECK2-NEXT: store i64 [[TMP95]], i64* [[TMP120]], align 8 // CHECK2-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4
// CHECK2-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 3 // CHECK2-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to [5 x [10 x double]]**
// CHECK2-NEXT: store i8* null, i8** [[TMP121]], align 8 // CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP121]], align 8
// CHECK2-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4 // CHECK2-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 4
// CHECK2-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to [5 x [10 x double]]** // CHECK2-NEXT: store i8* null, i8** [[TMP122]], align 8
// CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP123]], align 8 // CHECK2-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5
// CHECK2-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4 // CHECK2-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i64*
// CHECK2-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to [5 x [10 x double]]** // CHECK2-NEXT: store i64 5, i64* [[TMP124]], align 8
// CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP125]], align 8 // CHECK2-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5
// CHECK2-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK2-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64*
// CHECK2-NEXT: store i64 400, i64* [[TMP126]], align 8 // CHECK2-NEXT: store i64 5, i64* [[TMP126]], align 8
// CHECK2-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 4 // CHECK2-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 5
// CHECK2-NEXT: store i8* null, i8** [[TMP127]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP127]], align 8
// CHECK2-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5 // CHECK2-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6
// CHECK2-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i64* // CHECK2-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i64*
// CHECK2-NEXT: store i64 5, i64* [[TMP129]], align 8 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP129]], align 8
// CHECK2-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5 // CHECK2-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6
// CHECK2-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i64* // CHECK2-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i64*
// CHECK2-NEXT: store i64 5, i64* [[TMP131]], align 8 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP131]], align 8
// CHECK2-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK2-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 6
// CHECK2-NEXT: store i64 8, i64* [[TMP132]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP132]], align 8
// CHECK2-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 5 // CHECK2-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7
// CHECK2-NEXT: store i8* null, i8** [[TMP133]], align 8 // CHECK2-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double**
// CHECK2-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6 // CHECK2-NEXT: store double* [[VLA1]], double** [[TMP134]], align 8
// CHECK2-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i64* // CHECK2-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7
// CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP135]], align 8 // CHECK2-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to double**
// CHECK2-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6 // CHECK2-NEXT: store double* [[VLA1]], double** [[TMP136]], align 8
// CHECK2-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i64* // CHECK2-NEXT: store i64 [[TMP97]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 7), align 8
// CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP137]], align 8 // CHECK2-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 7
// CHECK2-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK2-NEXT: store i8* null, i8** [[TMP137]], align 8
// CHECK2-NEXT: store i64 8, i64* [[TMP138]], align 8 // CHECK2-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8
// CHECK2-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 6 // CHECK2-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to %struct.TT**
// CHECK2-NEXT: store i8* null, i8** [[TMP139]], align 8 // CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP139]], align 8
// CHECK2-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7 // CHECK2-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8
// CHECK2-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to double** // CHECK2-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to %struct.TT**
// CHECK2-NEXT: store double* [[VLA1]], double** [[TMP141]], align 8 // CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP141]], align 8
// CHECK2-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7 // CHECK2-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 8
// CHECK2-NEXT: [[TMP143:%.*]] = bitcast i8** [[TMP142]] to double** // CHECK2-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK2-NEXT: store double* [[VLA1]], double** [[TMP143]], align 8 // CHECK2-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9
// CHECK2-NEXT: [[TMP144:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK2-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i64*
// CHECK2-NEXT: store i64 [[TMP97]], i64* [[TMP144]], align 8 // CHECK2-NEXT: store i64 [[TMP93]], i64* [[TMP144]], align 8
// CHECK2-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 7 // CHECK2-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9
// CHECK2-NEXT: store i8* null, i8** [[TMP145]], align 8 // CHECK2-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i64*
// CHECK2-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8 // CHECK2-NEXT: store i64 [[TMP93]], i64* [[TMP146]], align 8
// CHECK2-NEXT: [[TMP147:%.*]] = bitcast i8** [[TMP146]] to %struct.TT** // CHECK2-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 9
// CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP147]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP147]], align 8
// CHECK2-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8 // CHECK2-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK2-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to %struct.TT** // CHECK2-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP149]], align 8
// CHECK2-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK2-NEXT: store i64 16, i64* [[TMP150]], align 8
// CHECK2-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 8
// CHECK2-NEXT: store i8* null, i8** [[TMP151]], align 8
// CHECK2-NEXT: [[TMP152:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9
// CHECK2-NEXT: [[TMP153:%.*]] = bitcast i8** [[TMP152]] to i64*
// CHECK2-NEXT: store i64 [[TMP93]], i64* [[TMP153]], align 8
// CHECK2-NEXT: [[TMP154:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9
// CHECK2-NEXT: [[TMP155:%.*]] = bitcast i8** [[TMP154]] to i64*
// CHECK2-NEXT: store i64 [[TMP93]], i64* [[TMP155]], align 8
// CHECK2-NEXT: [[TMP156:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK2-NEXT: store i64 4, i64* [[TMP156]], align 8
// CHECK2-NEXT: [[TMP157:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 9
// CHECK2-NEXT: store i8* null, i8** [[TMP157]], align 8
// CHECK2-NEXT: [[TMP158:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK2-NEXT: [[TMP159:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK2-NEXT: [[TMP160:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK2-NEXT: [[TMP161:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP158]], i8** [[TMP159]], i64* [[TMP160]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP148]], i8** [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP162:%.*]] = icmp ne i32 [[TMP161]], 0 // CHECK2-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK2-NEXT: br i1 [[TMP162]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]] // CHECK2-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK2: omp_offload.failed33: // CHECK2: omp_offload.failed33:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT34]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK2: omp_offload.cont34: // CHECK2: omp_offload.cont34:
// CHECK2-NEXT: br label [[OMP_IF_END36:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END36:%.*]]
// CHECK2: omp_if.else35: // CHECK2: omp_if.else35:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]]
// CHECK2-NEXT: br label [[OMP_IF_END36]] // CHECK2-NEXT: br label [[OMP_IF_END36]]
// CHECK2: omp_if.end36: // CHECK2: omp_if.end36:
// CHECK2-NEXT: [[TMP163:%.*]] = load i32, i32* [[A]], align 4 // CHECK2-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4
// CHECK2-NEXT: [[TMP164:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP164]]) // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP153]])
// CHECK2-NEXT: ret i32 [[TMP163]] // CHECK2-NEXT: ret i32 [[TMP152]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK2-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK2-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 678 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK2-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK2-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK2-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 15 Lines
// CHECK2-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK2-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK2-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK2-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK2-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK2-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK2-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK2-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK2-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK2-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK2-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK2-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK2-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK2-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK2-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK2-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK2-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK2-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK2-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK2-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK2-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK2-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK2-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK2-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK2-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK2-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 8
// CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK2-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK2-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK2-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8
// CHECK2-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK2-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK2-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK2-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK2-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK2-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK2-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK2-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]]
// CHECK2-NEXT: br label [[OMP_IF_END]] // CHECK2-NEXT: br label [[OMP_IF_END]]
// CHECK2: omp_if.end: // CHECK2: omp_if.end:
// CHECK2-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK2-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK2-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK2-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK2-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK2-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK2-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK2-NEXT: ret i32 [[ADD4]] // CHECK2-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK2-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK2-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1
// CHECK2-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK2-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK2-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1 // CHECK2-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1
// CHECK2-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK2-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK2-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK2-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK2-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1 // CHECK2-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1
// CHECK2-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64 // CHECK2-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK2-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK2-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK2-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK2-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK2-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK2-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK2-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
Show All 22 Lines
// CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2 // CHECK2-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2
// CHECK2-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK2-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK2-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2 // CHECK2-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2
// CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK2-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK2-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK2-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK2-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK2-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1 // CHECK2-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1
// CHECK2-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK2-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 129 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[_TMP15:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[_TMP15:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A_CASTED19:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_CASTED19:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED20:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED20:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [10 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [10 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [10 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [10 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [10 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [10 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK3-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[_TMP26:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[A]], align 4 // CHECK3-NEXT: store i32 0, i32* [[A]], align 4
// CHECK3-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK3-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK3-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
▲ Show 20 Lines • Show All 162 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP96:%.*]] = mul nuw i32 [[TMP95]], 8 // CHECK3-NEXT: [[TMP96:%.*]] = mul nuw i32 [[TMP95]], 8
// CHECK3-NEXT: [[TMP97:%.*]] = sext i32 [[TMP96]] to i64 // CHECK3-NEXT: [[TMP97:%.*]] = sext i32 [[TMP96]] to i64
// CHECK3-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0 // CHECK3-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK3-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32* // CHECK3-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32*
// CHECK3-NEXT: store i32 [[TMP89]], i32* [[TMP99]], align 4 // CHECK3-NEXT: store i32 [[TMP89]], i32* [[TMP99]], align 4
// CHECK3-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0 // CHECK3-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK3-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i32* // CHECK3-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i32*
// CHECK3-NEXT: store i32 [[TMP89]], i32* [[TMP101]], align 4 // CHECK3-NEXT: store i32 [[TMP89]], i32* [[TMP101]], align 4
// CHECK3-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK3-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 0
// CHECK3-NEXT: store i64 4, i64* [[TMP102]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP102]], align 4
// CHECK3-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 0 // CHECK3-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP103]], align 4 // CHECK3-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]**
// CHECK3-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1 // CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 4
// CHECK3-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to [10 x float]** // CHECK3-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 1
// CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP105]], align 4 // CHECK3-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to [10 x float]**
// CHECK3-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 1 // CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP106]], align 4
// CHECK3-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to [10 x float]** // CHECK3-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 1
// CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP107]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP107]], align 4
// CHECK3-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK3-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 2
// CHECK3-NEXT: store i64 40, i64* [[TMP108]], align 4 // CHECK3-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to i32*
// CHECK3-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP109]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP109]], align 4 // CHECK3-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 2
// CHECK3-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 2
// CHECK3-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i32* // CHECK3-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP111]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP111]], align 4
// CHECK3-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 2 // CHECK3-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 2
// CHECK3-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i32* // CHECK3-NEXT: store i8* null, i8** [[TMP112]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP113]], align 4 // CHECK3-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 3
// CHECK3-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK3-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float**
// CHECK3-NEXT: store i64 4, i64* [[TMP114]], align 4 // CHECK3-NEXT: store float* [[VLA]], float** [[TMP114]], align 4
// CHECK3-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 2 // CHECK3-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP115]], align 4 // CHECK3-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float**
// CHECK3-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 3 // CHECK3-NEXT: store float* [[VLA]], float** [[TMP116]], align 4
// CHECK3-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to float** // CHECK3-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 3), align 4
// CHECK3-NEXT: store float* [[VLA]], float** [[TMP117]], align 4 // CHECK3-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 3
// CHECK3-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 3 // CHECK3-NEXT: store i8* null, i8** [[TMP117]], align 4
// CHECK3-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to float** // CHECK3-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 4
// CHECK3-NEXT: store float* [[VLA]], float** [[TMP119]], align 4 // CHECK3-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to [5 x [10 x double]]**
// CHECK3-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP119]], align 4
// CHECK3-NEXT: store i64 [[TMP94]], i64* [[TMP120]], align 4 // CHECK3-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 4
// CHECK3-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 3 // CHECK3-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to [5 x [10 x double]]**
// CHECK3-NEXT: store i8* null, i8** [[TMP121]], align 4 // CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP121]], align 4
// CHECK3-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 4 // CHECK3-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 4
// CHECK3-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to [5 x [10 x double]]** // CHECK3-NEXT: store i8* null, i8** [[TMP122]], align 4
// CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP123]], align 4 // CHECK3-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 5
// CHECK3-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 4 // CHECK3-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i32*
// CHECK3-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to [5 x [10 x double]]** // CHECK3-NEXT: store i32 5, i32* [[TMP124]], align 4
// CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP125]], align 4 // CHECK3-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 5
// CHECK3-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK3-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32*
// CHECK3-NEXT: store i64 400, i64* [[TMP126]], align 4 // CHECK3-NEXT: store i32 5, i32* [[TMP126]], align 4
// CHECK3-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 4 // CHECK3-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 5
// CHECK3-NEXT: store i8* null, i8** [[TMP127]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP127]], align 4
// CHECK3-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 5 // CHECK3-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 6
// CHECK3-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i32* // CHECK3-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i32*
// CHECK3-NEXT: store i32 5, i32* [[TMP129]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP129]], align 4
// CHECK3-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 5 // CHECK3-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 6
// CHECK3-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i32* // CHECK3-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i32*
// CHECK3-NEXT: store i32 5, i32* [[TMP131]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP131]], align 4
// CHECK3-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK3-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 6
// CHECK3-NEXT: store i64 4, i64* [[TMP132]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP132]], align 4
// CHECK3-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 5 // CHECK3-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 7
// CHECK3-NEXT: store i8* null, i8** [[TMP133]], align 4 // CHECK3-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double**
// CHECK3-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 6 // CHECK3-NEXT: store double* [[VLA1]], double** [[TMP134]], align 4
// CHECK3-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i32* // CHECK3-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 7
// CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP135]], align 4 // CHECK3-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to double**
// CHECK3-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 6 // CHECK3-NEXT: store double* [[VLA1]], double** [[TMP136]], align 4
// CHECK3-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i32* // CHECK3-NEXT: store i64 [[TMP97]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 7), align 4
// CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP137]], align 4 // CHECK3-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 7
// CHECK3-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK3-NEXT: store i8* null, i8** [[TMP137]], align 4
// CHECK3-NEXT: store i64 4, i64* [[TMP138]], align 4 // CHECK3-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 8
// CHECK3-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 6 // CHECK3-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to %struct.TT**
// CHECK3-NEXT: store i8* null, i8** [[TMP139]], align 4 // CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP139]], align 4
// CHECK3-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 7 // CHECK3-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 8
// CHECK3-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to double** // CHECK3-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to %struct.TT**
// CHECK3-NEXT: store double* [[VLA1]], double** [[TMP141]], align 4 // CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP141]], align 4
// CHECK3-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 7 // CHECK3-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 8
// CHECK3-NEXT: [[TMP143:%.*]] = bitcast i8** [[TMP142]] to double** // CHECK3-NEXT: store i8* null, i8** [[TMP142]], align 4
// CHECK3-NEXT: store double* [[VLA1]], double** [[TMP143]], align 4 // CHECK3-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 9
// CHECK3-NEXT: [[TMP144:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK3-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i32*
// CHECK3-NEXT: store i64 [[TMP97]], i64* [[TMP144]], align 4 // CHECK3-NEXT: store i32 [[TMP91]], i32* [[TMP144]], align 4
// CHECK3-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 7 // CHECK3-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 9
// CHECK3-NEXT: store i8* null, i8** [[TMP145]], align 4 // CHECK3-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i32*
// CHECK3-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 8 // CHECK3-NEXT: store i32 [[TMP91]], i32* [[TMP146]], align 4
// CHECK3-NEXT: [[TMP147:%.*]] = bitcast i8** [[TMP146]] to %struct.TT** // CHECK3-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 9
// CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP147]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP147]], align 4
// CHECK3-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 8 // CHECK3-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK3-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to %struct.TT** // CHECK3-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP149]], align 4
// CHECK3-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK3-NEXT: store i64 12, i64* [[TMP150]], align 4
// CHECK3-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 8
// CHECK3-NEXT: store i8* null, i8** [[TMP151]], align 4
// CHECK3-NEXT: [[TMP152:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 9
// CHECK3-NEXT: [[TMP153:%.*]] = bitcast i8** [[TMP152]] to i32*
// CHECK3-NEXT: store i32 [[TMP91]], i32* [[TMP153]], align 4
// CHECK3-NEXT: [[TMP154:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 9
// CHECK3-NEXT: [[TMP155:%.*]] = bitcast i8** [[TMP154]] to i32*
// CHECK3-NEXT: store i32 [[TMP91]], i32* [[TMP155]], align 4
// CHECK3-NEXT: [[TMP156:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK3-NEXT: store i64 4, i64* [[TMP156]], align 4
// CHECK3-NEXT: [[TMP157:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 9
// CHECK3-NEXT: store i8* null, i8** [[TMP157]], align 4
// CHECK3-NEXT: [[TMP158:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK3-NEXT: [[TMP159:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK3-NEXT: [[TMP160:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK3-NEXT: [[TMP161:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP158]], i8** [[TMP159]], i64* [[TMP160]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK3-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP148]], i8** [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK3-NEXT: [[TMP162:%.*]] = icmp ne i32 [[TMP161]], 0 // CHECK3-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK3-NEXT: br i1 [[TMP162]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]] // CHECK3-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK3: omp_offload.failed27: // CHECK3: omp_offload.failed27:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT28]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK3: omp_offload.cont28: // CHECK3: omp_offload.cont28:
// CHECK3-NEXT: br label [[OMP_IF_END30:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END30:%.*]]
// CHECK3: omp_if.else29: // CHECK3: omp_if.else29:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]]
// CHECK3-NEXT: br label [[OMP_IF_END30]] // CHECK3-NEXT: br label [[OMP_IF_END30]]
// CHECK3: omp_if.end30: // CHECK3: omp_if.end30:
// CHECK3-NEXT: [[TMP163:%.*]] = load i32, i32* [[A]], align 4 // CHECK3-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4
// CHECK3-NEXT: [[TMP164:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP164]]) // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP153]])
// CHECK3-NEXT: ret i32 [[TMP163]] // CHECK3-NEXT: ret i32 [[TMP152]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK3-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK3-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 662 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK3-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK3-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 14 Lines
// CHECK3-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK3-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK3-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK3-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK3-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK3-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK3-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK3-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK3-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK3-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK3-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK3-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK3-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK3-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK3-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK3-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK3-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK3-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK3-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK3-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK3-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 4
// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK3-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4
// CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK3-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK3-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK3-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK3-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK3-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK3-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK3-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]]
// CHECK3-NEXT: br label [[OMP_IF_END]] // CHECK3-NEXT: br label [[OMP_IF_END]]
// CHECK3: omp_if.end: // CHECK3: omp_if.end:
// CHECK3-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK3-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK3-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK3-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK3-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK3-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK3-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK3-NEXT: ret i32 [[ADD3]] // CHECK3-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK3-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK3-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1 // CHECK3-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1
// CHECK3-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK3-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK3-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1 // CHECK3-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1
// CHECK3-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK3-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK3-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK3-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK3-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1 // CHECK3-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1
// CHECK3-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64 // CHECK3-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64
// CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK3-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK3-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK3-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK3-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK3-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
▲ Show 20 Lines • Show All 53 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
Show All 20 Lines
// CHECK3-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK3-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 105 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK3-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK3-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK3-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2 // CHECK3-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK3-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK3-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK3-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK3-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1 // CHECK3-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 125 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK3-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK3-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 109 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[_TMP15:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[_TMP15:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[A_CASTED19:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_CASTED19:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTCAPTURE_EXPR__CASTED20:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__CASTED20:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [10 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [10 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [10 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [10 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [10 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [10 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK4-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[_TMP26:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK4-NEXT: store i32 0, i32* [[A]], align 4 // CHECK4-NEXT: store i32 0, i32* [[A]], align 4
// CHECK4-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK4-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK4-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
▲ Show 20 Lines • Show All 162 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP96:%.*]] = mul nuw i32 [[TMP95]], 8 // CHECK4-NEXT: [[TMP96:%.*]] = mul nuw i32 [[TMP95]], 8
// CHECK4-NEXT: [[TMP97:%.*]] = sext i32 [[TMP96]] to i64 // CHECK4-NEXT: [[TMP97:%.*]] = sext i32 [[TMP96]] to i64
// CHECK4-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0 // CHECK4-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK4-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32* // CHECK4-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32*
// CHECK4-NEXT: store i32 [[TMP89]], i32* [[TMP99]], align 4 // CHECK4-NEXT: store i32 [[TMP89]], i32* [[TMP99]], align 4
// CHECK4-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0 // CHECK4-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK4-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i32* // CHECK4-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i32*
// CHECK4-NEXT: store i32 [[TMP89]], i32* [[TMP101]], align 4 // CHECK4-NEXT: store i32 [[TMP89]], i32* [[TMP101]], align 4
// CHECK4-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK4-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 0
// CHECK4-NEXT: store i64 4, i64* [[TMP102]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP102]], align 4
// CHECK4-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 0 // CHECK4-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1
// CHECK4-NEXT: store i8* null, i8** [[TMP103]], align 4 // CHECK4-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]**
// CHECK4-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1 // CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 4
// CHECK4-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to [10 x float]** // CHECK4-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 1
// CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP105]], align 4 // CHECK4-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to [10 x float]**
// CHECK4-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 1 // CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP106]], align 4
// CHECK4-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to [10 x float]** // CHECK4-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 1
// CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP107]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP107]], align 4
// CHECK4-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK4-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 2
// CHECK4-NEXT: store i64 40, i64* [[TMP108]], align 4 // CHECK4-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to i32*
// CHECK4-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP109]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP109]], align 4 // CHECK4-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 2
// CHECK4-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 2
// CHECK4-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i32* // CHECK4-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i32*
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP111]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP111]], align 4
// CHECK4-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 2 // CHECK4-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 2
// CHECK4-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i32* // CHECK4-NEXT: store i8* null, i8** [[TMP112]], align 4
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP113]], align 4 // CHECK4-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 3
// CHECK4-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK4-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float**
// CHECK4-NEXT: store i64 4, i64* [[TMP114]], align 4 // CHECK4-NEXT: store float* [[VLA]], float** [[TMP114]], align 4
// CHECK4-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 2 // CHECK4-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP115]], align 4 // CHECK4-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float**
// CHECK4-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 3 // CHECK4-NEXT: store float* [[VLA]], float** [[TMP116]], align 4
// CHECK4-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to float** // CHECK4-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 3), align 4
// CHECK4-NEXT: store float* [[VLA]], float** [[TMP117]], align 4 // CHECK4-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 3
// CHECK4-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 3 // CHECK4-NEXT: store i8* null, i8** [[TMP117]], align 4
// CHECK4-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to float** // CHECK4-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 4
// CHECK4-NEXT: store float* [[VLA]], float** [[TMP119]], align 4 // CHECK4-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to [5 x [10 x double]]**
// CHECK4-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP119]], align 4
// CHECK4-NEXT: store i64 [[TMP94]], i64* [[TMP120]], align 4 // CHECK4-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 4
// CHECK4-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 3 // CHECK4-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to [5 x [10 x double]]**
// CHECK4-NEXT: store i8* null, i8** [[TMP121]], align 4 // CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP121]], align 4
// CHECK4-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 4 // CHECK4-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 4
// CHECK4-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to [5 x [10 x double]]** // CHECK4-NEXT: store i8* null, i8** [[TMP122]], align 4
// CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP123]], align 4 // CHECK4-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 5
// CHECK4-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 4 // CHECK4-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i32*
// CHECK4-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to [5 x [10 x double]]** // CHECK4-NEXT: store i32 5, i32* [[TMP124]], align 4
// CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP125]], align 4 // CHECK4-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 5
// CHECK4-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK4-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32*
// CHECK4-NEXT: store i64 400, i64* [[TMP126]], align 4 // CHECK4-NEXT: store i32 5, i32* [[TMP126]], align 4
// CHECK4-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 4 // CHECK4-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 5
// CHECK4-NEXT: store i8* null, i8** [[TMP127]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP127]], align 4
// CHECK4-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 5 // CHECK4-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 6
// CHECK4-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i32* // CHECK4-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i32*
// CHECK4-NEXT: store i32 5, i32* [[TMP129]], align 4 // CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP129]], align 4
// CHECK4-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 5 // CHECK4-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 6
// CHECK4-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i32* // CHECK4-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i32*
// CHECK4-NEXT: store i32 5, i32* [[TMP131]], align 4 // CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP131]], align 4
// CHECK4-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK4-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 6
// CHECK4-NEXT: store i64 4, i64* [[TMP132]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP132]], align 4
// CHECK4-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 5 // CHECK4-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 7
// CHECK4-NEXT: store i8* null, i8** [[TMP133]], align 4 // CHECK4-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double**
// CHECK4-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 6 // CHECK4-NEXT: store double* [[VLA1]], double** [[TMP134]], align 4
// CHECK4-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i32* // CHECK4-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 7
// CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP135]], align 4 // CHECK4-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to double**
// CHECK4-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 6 // CHECK4-NEXT: store double* [[VLA1]], double** [[TMP136]], align 4
// CHECK4-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i32* // CHECK4-NEXT: store i64 [[TMP97]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 7), align 4
// CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP137]], align 4 // CHECK4-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 7
// CHECK4-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK4-NEXT: store i8* null, i8** [[TMP137]], align 4
// CHECK4-NEXT: store i64 4, i64* [[TMP138]], align 4 // CHECK4-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 8
// CHECK4-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 6 // CHECK4-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to %struct.TT**
// CHECK4-NEXT: store i8* null, i8** [[TMP139]], align 4 // CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP139]], align 4
// CHECK4-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 7 // CHECK4-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 8
// CHECK4-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to double** // CHECK4-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to %struct.TT**
// CHECK4-NEXT: store double* [[VLA1]], double** [[TMP141]], align 4 // CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP141]], align 4
// CHECK4-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 7 // CHECK4-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 8
// CHECK4-NEXT: [[TMP143:%.*]] = bitcast i8** [[TMP142]] to double** // CHECK4-NEXT: store i8* null, i8** [[TMP142]], align 4
// CHECK4-NEXT: store double* [[VLA1]], double** [[TMP143]], align 4 // CHECK4-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 9
// CHECK4-NEXT: [[TMP144:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK4-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i32*
// CHECK4-NEXT: store i64 [[TMP97]], i64* [[TMP144]], align 4 // CHECK4-NEXT: store i32 [[TMP91]], i32* [[TMP144]], align 4
// CHECK4-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 7 // CHECK4-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 9
// CHECK4-NEXT: store i8* null, i8** [[TMP145]], align 4 // CHECK4-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i32*
// CHECK4-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 8 // CHECK4-NEXT: store i32 [[TMP91]], i32* [[TMP146]], align 4
// CHECK4-NEXT: [[TMP147:%.*]] = bitcast i8** [[TMP146]] to %struct.TT** // CHECK4-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 9
// CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP147]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP147]], align 4
// CHECK4-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 8 // CHECK4-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK4-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to %struct.TT** // CHECK4-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP149]], align 4
// CHECK4-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK4-NEXT: store i64 12, i64* [[TMP150]], align 4
// CHECK4-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 8
// CHECK4-NEXT: store i8* null, i8** [[TMP151]], align 4
// CHECK4-NEXT: [[TMP152:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 9
// CHECK4-NEXT: [[TMP153:%.*]] = bitcast i8** [[TMP152]] to i32*
// CHECK4-NEXT: store i32 [[TMP91]], i32* [[TMP153]], align 4
// CHECK4-NEXT: [[TMP154:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 9
// CHECK4-NEXT: [[TMP155:%.*]] = bitcast i8** [[TMP154]] to i32*
// CHECK4-NEXT: store i32 [[TMP91]], i32* [[TMP155]], align 4
// CHECK4-NEXT: [[TMP156:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK4-NEXT: store i64 4, i64* [[TMP156]], align 4
// CHECK4-NEXT: [[TMP157:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 9
// CHECK4-NEXT: store i8* null, i8** [[TMP157]], align 4
// CHECK4-NEXT: [[TMP158:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK4-NEXT: [[TMP159:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK4-NEXT: [[TMP160:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK4-NEXT: [[TMP161:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP158]], i8** [[TMP159]], i64* [[TMP160]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK4-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP148]], i8** [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK4-NEXT: [[TMP162:%.*]] = icmp ne i32 [[TMP161]], 0 // CHECK4-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK4-NEXT: br i1 [[TMP162]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]] // CHECK4-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK4: omp_offload.failed27: // CHECK4: omp_offload.failed27:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT28]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK4: omp_offload.cont28: // CHECK4: omp_offload.cont28:
// CHECK4-NEXT: br label [[OMP_IF_END30:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END30:%.*]]
// CHECK4: omp_if.else29: // CHECK4: omp_if.else29:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]]
// CHECK4-NEXT: br label [[OMP_IF_END30]] // CHECK4-NEXT: br label [[OMP_IF_END30]]
// CHECK4: omp_if.end30: // CHECK4: omp_if.end30:
// CHECK4-NEXT: [[TMP163:%.*]] = load i32, i32* [[A]], align 4 // CHECK4-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4
// CHECK4-NEXT: [[TMP164:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP164]]) // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP153]])
// CHECK4-NEXT: ret i32 [[TMP163]] // CHECK4-NEXT: ret i32 [[TMP152]]
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK4-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK4-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 662 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK4-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK4-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK4-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 14 Lines
// CHECK4-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK4-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK4-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK4-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK4-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK4-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK4-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK4-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK4-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK4-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK4-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK4-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK4-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK4-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK4-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK4-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK4-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK4-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK4-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK4-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK4-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK4-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK4-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK4-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK4-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK4-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK4-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK4-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK4-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK4-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK4-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK4-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK4-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK4-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK4-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK4-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK4-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK4-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 4
// CHECK4-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK4-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK4-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK4-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK4-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK4-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4
// CHECK4-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK4-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK4-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK4-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK4-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK4-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK4-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK4-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK4-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK4-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]]
// CHECK4-NEXT: br label [[OMP_IF_END]] // CHECK4-NEXT: br label [[OMP_IF_END]]
// CHECK4: omp_if.end: // CHECK4: omp_if.end:
// CHECK4-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK4-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK4-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK4-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK4-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK4-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK4-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK4-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK4-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK4-NEXT: ret i32 [[ADD3]] // CHECK4-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK4-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK4-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK4-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1 // CHECK4-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1
// CHECK4-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK4-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK4-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1 // CHECK4-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1
// CHECK4-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK4-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK4-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK4-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK4-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1 // CHECK4-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1
// CHECK4-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64 // CHECK4-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64
// CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK4-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK4-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK4-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK4-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK4-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
▲ Show 20 Lines • Show All 53 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK4-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK4-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK4-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK4-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK4-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
Show All 20 Lines
// CHECK4-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK4-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK4-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK4-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 105 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK4-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK4-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK4-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK4-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2 // CHECK4-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2
// CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK4-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK4-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK4-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK4-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK4-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1 // CHECK4-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1
// CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 125 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK4-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK4-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 3,455 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8
// CHECK17-NEXT: [[_TMP19:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[_TMP19:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[A_CASTED23:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[A_CASTED23:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED25:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__CASTED25:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [10 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [10 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [10 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [10 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [10 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [10 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK17-NEXT: [[_TMP32:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[_TMP32:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK17-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK17-NEXT: store i32 0, i32* [[A]], align 4 // CHECK17-NEXT: store i32 0, i32* [[A]], align 4
// CHECK17-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK17-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK17-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK17-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK17-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
▲ Show 20 Lines • Show All 168 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[TMP96:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK17-NEXT: [[TMP96:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK17-NEXT: [[TMP97:%.*]] = mul nuw i64 [[TMP96]], 8 // CHECK17-NEXT: [[TMP97:%.*]] = mul nuw i64 [[TMP96]], 8
// CHECK17-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0 // CHECK17-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK17-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64* // CHECK17-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64*
// CHECK17-NEXT: store i64 [[TMP91]], i64* [[TMP99]], align 8 // CHECK17-NEXT: store i64 [[TMP91]], i64* [[TMP99]], align 8
// CHECK17-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0 // CHECK17-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK17-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64* // CHECK17-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64*
// CHECK17-NEXT: store i64 [[TMP91]], i64* [[TMP101]], align 8 // CHECK17-NEXT: store i64 [[TMP91]], i64* [[TMP101]], align 8
// CHECK17-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK17-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 0
// CHECK17-NEXT: store i64 4, i64* [[TMP102]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK17-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 0 // CHECK17-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1
// CHECK17-NEXT: store i8* null, i8** [[TMP103]], align 8 // CHECK17-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]**
// CHECK17-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1 // CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 8
// CHECK17-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to [10 x float]** // CHECK17-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1
// CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP105]], align 8 // CHECK17-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to [10 x float]**
// CHECK17-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1 // CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP106]], align 8
// CHECK17-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to [10 x float]** // CHECK17-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 1
// CHECK17-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP107]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP107]], align 8
// CHECK17-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK17-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2
// CHECK17-NEXT: store i64 40, i64* [[TMP108]], align 8 // CHECK17-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to i64*
// CHECK17-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 1 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP109]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP109]], align 8 // CHECK17-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2
// CHECK17-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2
// CHECK17-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i64* // CHECK17-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i64*
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP111]], align 8 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP111]], align 8
// CHECK17-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2 // CHECK17-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 2
// CHECK17-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i64* // CHECK17-NEXT: store i8* null, i8** [[TMP112]], align 8
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP113]], align 8 // CHECK17-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3
// CHECK17-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK17-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float**
// CHECK17-NEXT: store i64 8, i64* [[TMP114]], align 8 // CHECK17-NEXT: store float* [[VLA]], float** [[TMP114]], align 8
// CHECK17-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 2 // CHECK17-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3
// CHECK17-NEXT: store i8* null, i8** [[TMP115]], align 8 // CHECK17-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float**
// CHECK17-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3 // CHECK17-NEXT: store float* [[VLA]], float** [[TMP116]], align 8
// CHECK17-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to float** // CHECK17-NEXT: store i64 [[TMP95]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 3), align 8
// CHECK17-NEXT: store float* [[VLA]], float** [[TMP117]], align 8 // CHECK17-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 3
// CHECK17-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3 // CHECK17-NEXT: store i8* null, i8** [[TMP117]], align 8
// CHECK17-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to float** // CHECK17-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4
// CHECK17-NEXT: store float* [[VLA]], float** [[TMP119]], align 8 // CHECK17-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to [5 x [10 x double]]**
// CHECK17-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP119]], align 8
// CHECK17-NEXT: store i64 [[TMP95]], i64* [[TMP120]], align 8 // CHECK17-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4
// CHECK17-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 3 // CHECK17-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to [5 x [10 x double]]**
// CHECK17-NEXT: store i8* null, i8** [[TMP121]], align 8 // CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP121]], align 8
// CHECK17-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4 // CHECK17-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 4
// CHECK17-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to [5 x [10 x double]]** // CHECK17-NEXT: store i8* null, i8** [[TMP122]], align 8
// CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP123]], align 8 // CHECK17-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5
// CHECK17-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4 // CHECK17-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i64*
// CHECK17-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to [5 x [10 x double]]** // CHECK17-NEXT: store i64 5, i64* [[TMP124]], align 8
// CHECK17-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP125]], align 8 // CHECK17-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5
// CHECK17-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK17-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64*
// CHECK17-NEXT: store i64 400, i64* [[TMP126]], align 8 // CHECK17-NEXT: store i64 5, i64* [[TMP126]], align 8
// CHECK17-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 4 // CHECK17-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 5
// CHECK17-NEXT: store i8* null, i8** [[TMP127]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP127]], align 8
// CHECK17-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5 // CHECK17-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6
// CHECK17-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i64* // CHECK17-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i64*
// CHECK17-NEXT: store i64 5, i64* [[TMP129]], align 8 // CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP129]], align 8
// CHECK17-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5 // CHECK17-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6
// CHECK17-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i64* // CHECK17-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i64*
// CHECK17-NEXT: store i64 5, i64* [[TMP131]], align 8 // CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP131]], align 8
// CHECK17-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK17-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 6
// CHECK17-NEXT: store i64 8, i64* [[TMP132]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP132]], align 8
// CHECK17-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 5 // CHECK17-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7
// CHECK17-NEXT: store i8* null, i8** [[TMP133]], align 8 // CHECK17-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double**
// CHECK17-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6 // CHECK17-NEXT: store double* [[VLA1]], double** [[TMP134]], align 8
// CHECK17-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i64* // CHECK17-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7
// CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP135]], align 8 // CHECK17-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to double**
// CHECK17-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6 // CHECK17-NEXT: store double* [[VLA1]], double** [[TMP136]], align 8
// CHECK17-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i64* // CHECK17-NEXT: store i64 [[TMP97]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 7), align 8
// CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP137]], align 8 // CHECK17-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 7
// CHECK17-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK17-NEXT: store i8* null, i8** [[TMP137]], align 8
// CHECK17-NEXT: store i64 8, i64* [[TMP138]], align 8 // CHECK17-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8
// CHECK17-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 6 // CHECK17-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to %struct.TT**
// CHECK17-NEXT: store i8* null, i8** [[TMP139]], align 8 // CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP139]], align 8
// CHECK17-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7 // CHECK17-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8
// CHECK17-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to double** // CHECK17-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to %struct.TT**
// CHECK17-NEXT: store double* [[VLA1]], double** [[TMP141]], align 8 // CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP141]], align 8
// CHECK17-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7 // CHECK17-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 8
// CHECK17-NEXT: [[TMP143:%.*]] = bitcast i8** [[TMP142]] to double** // CHECK17-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK17-NEXT: store double* [[VLA1]], double** [[TMP143]], align 8 // CHECK17-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9
// CHECK17-NEXT: [[TMP144:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK17-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i64*
// CHECK17-NEXT: store i64 [[TMP97]], i64* [[TMP144]], align 8 // CHECK17-NEXT: store i64 [[TMP93]], i64* [[TMP144]], align 8
// CHECK17-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 7 // CHECK17-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9
// CHECK17-NEXT: store i8* null, i8** [[TMP145]], align 8 // CHECK17-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i64*
// CHECK17-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8 // CHECK17-NEXT: store i64 [[TMP93]], i64* [[TMP146]], align 8
// CHECK17-NEXT: [[TMP147:%.*]] = bitcast i8** [[TMP146]] to %struct.TT** // CHECK17-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 9
// CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP147]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP147]], align 8
// CHECK17-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8 // CHECK17-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK17-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to %struct.TT** // CHECK17-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK17-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP149]], align 8
// CHECK17-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK17-NEXT: store i64 16, i64* [[TMP150]], align 8
// CHECK17-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 8
// CHECK17-NEXT: store i8* null, i8** [[TMP151]], align 8
// CHECK17-NEXT: [[TMP152:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9
// CHECK17-NEXT: [[TMP153:%.*]] = bitcast i8** [[TMP152]] to i64*
// CHECK17-NEXT: store i64 [[TMP93]], i64* [[TMP153]], align 8
// CHECK17-NEXT: [[TMP154:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9
// CHECK17-NEXT: [[TMP155:%.*]] = bitcast i8** [[TMP154]] to i64*
// CHECK17-NEXT: store i64 [[TMP93]], i64* [[TMP155]], align 8
// CHECK17-NEXT: [[TMP156:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK17-NEXT: store i64 4, i64* [[TMP156]], align 8
// CHECK17-NEXT: [[TMP157:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 9
// CHECK17-NEXT: store i8* null, i8** [[TMP157]], align 8
// CHECK17-NEXT: [[TMP158:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK17-NEXT: [[TMP159:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK17-NEXT: [[TMP160:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK17-NEXT: [[TMP161:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP158]], i8** [[TMP159]], i64* [[TMP160]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK17-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP148]], i8** [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK17-NEXT: [[TMP162:%.*]] = icmp ne i32 [[TMP161]], 0 // CHECK17-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK17-NEXT: br i1 [[TMP162]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]] // CHECK17-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK17: omp_offload.failed33: // CHECK17: omp_offload.failed33:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT34]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK17: omp_offload.cont34: // CHECK17: omp_offload.cont34:
// CHECK17-NEXT: br label [[OMP_IF_END36:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END36:%.*]]
// CHECK17: omp_if.else35: // CHECK17: omp_if.else35:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]]
// CHECK17-NEXT: br label [[OMP_IF_END36]] // CHECK17-NEXT: br label [[OMP_IF_END36]]
// CHECK17: omp_if.end36: // CHECK17: omp_if.end36:
// CHECK17-NEXT: [[TMP163:%.*]] = load i32, i32* [[A]], align 4 // CHECK17-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4
// CHECK17-NEXT: [[TMP164:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK17-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP164]]) // CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP153]])
// CHECK17-NEXT: ret i32 [[TMP163]] // CHECK17-NEXT: ret i32 [[TMP152]]
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK17-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK17-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 678 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK17-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK17-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK17-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK17-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK17-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 15 Lines
// CHECK17-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK17-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK17-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK17-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK17-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK17-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK17-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK17-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK17-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK17-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK17-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK17-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK17-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK17-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK17-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK17-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK17-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK17-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK17-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK17-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK17-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK17-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK17-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK17-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK17-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK17-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK17-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK17-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK17-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK17-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK17-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK17-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK17-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK17-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK17-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK17-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK17-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK17-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 8
// CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK17-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK17-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK17-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK17-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK17-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8
// CHECK17-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK17-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK17-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK17-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK17-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK17-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK17-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK17-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK17-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK17-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK17-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: br label [[OMP_IF_END:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK17: omp_if.else: // CHECK17: omp_if.else:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]]
// CHECK17-NEXT: br label [[OMP_IF_END]] // CHECK17-NEXT: br label [[OMP_IF_END]]
// CHECK17: omp_if.end: // CHECK17: omp_if.end:
// CHECK17-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK17-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK17-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK17-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK17-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK17-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK17-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK17-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK17-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK17-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK17-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK17-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK17-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK17-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK17-NEXT: ret i32 [[ADD4]] // CHECK17-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK17-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK17-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK17-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK17-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1 // CHECK17-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1
// CHECK17-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK17-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK17-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1 // CHECK17-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1
// CHECK17-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK17-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK17-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK17-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK17-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1 // CHECK17-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1
// CHECK17-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64 // CHECK17-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64
// CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK17-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK17-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK17-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK17-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK17-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK17-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: br label [[OMP_IF_END:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK17: omp_if.else: // CHECK17: omp_if.else:
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK17-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK17-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK17-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK17-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK17-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK17-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK17-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK17-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: br label [[OMP_IF_END:%.*]] // CHECK17-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK17: omp_if.else: // CHECK17: omp_if.else:
Show All 22 Lines
// CHECK17-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK17-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK17-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK17-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK17-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK17-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK17-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK17-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK17-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] { // CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2 // CHECK17-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2
// CHECK17-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK17-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK17-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2 // CHECK17-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2
// CHECK17-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK17-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK17-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK17-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK17-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK17-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK17-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1 // CHECK17-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1
// CHECK17-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK17-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 129 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK17-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK17-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK17-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK17-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK17-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK17-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK17-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK17-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK17-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK17-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK17-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK17-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK17-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8
// CHECK18-NEXT: [[_TMP19:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[_TMP19:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[A_CASTED23:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[A_CASTED23:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTCAPTURE_EXPR__CASTED25:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[DOTCAPTURE_EXPR__CASTED25:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [10 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS29:%.*]] = alloca [10 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [10 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS30:%.*]] = alloca [10 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [10 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS31:%.*]] = alloca [10 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK18-NEXT: [[_TMP32:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[_TMP32:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK18-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK18-NEXT: store i32 0, i32* [[A]], align 4 // CHECK18-NEXT: store i32 0, i32* [[A]], align 4
// CHECK18-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK18-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK18-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK18-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK18-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
▲ Show 20 Lines • Show All 168 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[TMP96:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK18-NEXT: [[TMP96:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK18-NEXT: [[TMP97:%.*]] = mul nuw i64 [[TMP96]], 8 // CHECK18-NEXT: [[TMP97:%.*]] = mul nuw i64 [[TMP96]], 8
// CHECK18-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0 // CHECK18-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK18-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64* // CHECK18-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i64*
// CHECK18-NEXT: store i64 [[TMP91]], i64* [[TMP99]], align 8 // CHECK18-NEXT: store i64 [[TMP91]], i64* [[TMP99]], align 8
// CHECK18-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0 // CHECK18-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK18-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64* // CHECK18-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i64*
// CHECK18-NEXT: store i64 [[TMP91]], i64* [[TMP101]], align 8 // CHECK18-NEXT: store i64 [[TMP91]], i64* [[TMP101]], align 8
// CHECK18-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK18-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 0
// CHECK18-NEXT: store i64 4, i64* [[TMP102]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP102]], align 8
// CHECK18-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 0 // CHECK18-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1
// CHECK18-NEXT: store i8* null, i8** [[TMP103]], align 8 // CHECK18-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]**
// CHECK18-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 1 // CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 8
// CHECK18-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to [10 x float]** // CHECK18-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1
// CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP105]], align 8 // CHECK18-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to [10 x float]**
// CHECK18-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 1 // CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP106]], align 8
// CHECK18-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to [10 x float]** // CHECK18-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 1
// CHECK18-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP107]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP107]], align 8
// CHECK18-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK18-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2
// CHECK18-NEXT: store i64 40, i64* [[TMP108]], align 8 // CHECK18-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to i64*
// CHECK18-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 1 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP109]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP109]], align 8 // CHECK18-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2
// CHECK18-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 2
// CHECK18-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i64* // CHECK18-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i64*
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP111]], align 8 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP111]], align 8
// CHECK18-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 2 // CHECK18-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 2
// CHECK18-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i64* // CHECK18-NEXT: store i8* null, i8** [[TMP112]], align 8
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP113]], align 8 // CHECK18-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3
// CHECK18-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK18-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float**
// CHECK18-NEXT: store i64 8, i64* [[TMP114]], align 8 // CHECK18-NEXT: store float* [[VLA]], float** [[TMP114]], align 8
// CHECK18-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 2 // CHECK18-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3
// CHECK18-NEXT: store i8* null, i8** [[TMP115]], align 8 // CHECK18-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float**
// CHECK18-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 3 // CHECK18-NEXT: store float* [[VLA]], float** [[TMP116]], align 8
// CHECK18-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to float** // CHECK18-NEXT: store i64 [[TMP95]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 3), align 8
// CHECK18-NEXT: store float* [[VLA]], float** [[TMP117]], align 8 // CHECK18-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 3
// CHECK18-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 3 // CHECK18-NEXT: store i8* null, i8** [[TMP117]], align 8
// CHECK18-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to float** // CHECK18-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4
// CHECK18-NEXT: store float* [[VLA]], float** [[TMP119]], align 8 // CHECK18-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to [5 x [10 x double]]**
// CHECK18-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP119]], align 8
// CHECK18-NEXT: store i64 [[TMP95]], i64* [[TMP120]], align 8 // CHECK18-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4
// CHECK18-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 3 // CHECK18-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to [5 x [10 x double]]**
// CHECK18-NEXT: store i8* null, i8** [[TMP121]], align 8 // CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP121]], align 8
// CHECK18-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 4 // CHECK18-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 4
// CHECK18-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to [5 x [10 x double]]** // CHECK18-NEXT: store i8* null, i8** [[TMP122]], align 8
// CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP123]], align 8 // CHECK18-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5
// CHECK18-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 4 // CHECK18-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i64*
// CHECK18-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to [5 x [10 x double]]** // CHECK18-NEXT: store i64 5, i64* [[TMP124]], align 8
// CHECK18-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP125]], align 8 // CHECK18-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5
// CHECK18-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK18-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64*
// CHECK18-NEXT: store i64 400, i64* [[TMP126]], align 8 // CHECK18-NEXT: store i64 5, i64* [[TMP126]], align 8
// CHECK18-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 4 // CHECK18-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 5
// CHECK18-NEXT: store i8* null, i8** [[TMP127]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP127]], align 8
// CHECK18-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 5 // CHECK18-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6
// CHECK18-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i64* // CHECK18-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i64*
// CHECK18-NEXT: store i64 5, i64* [[TMP129]], align 8 // CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP129]], align 8
// CHECK18-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 5 // CHECK18-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6
// CHECK18-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i64* // CHECK18-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i64*
// CHECK18-NEXT: store i64 5, i64* [[TMP131]], align 8 // CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP131]], align 8
// CHECK18-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK18-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 6
// CHECK18-NEXT: store i64 8, i64* [[TMP132]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP132]], align 8
// CHECK18-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 5 // CHECK18-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7
// CHECK18-NEXT: store i8* null, i8** [[TMP133]], align 8 // CHECK18-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double**
// CHECK18-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 6 // CHECK18-NEXT: store double* [[VLA1]], double** [[TMP134]], align 8
// CHECK18-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i64* // CHECK18-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7
// CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP135]], align 8 // CHECK18-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to double**
// CHECK18-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 6 // CHECK18-NEXT: store double* [[VLA1]], double** [[TMP136]], align 8
// CHECK18-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i64* // CHECK18-NEXT: store i64 [[TMP97]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 7), align 8
// CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP137]], align 8 // CHECK18-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 7
// CHECK18-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK18-NEXT: store i8* null, i8** [[TMP137]], align 8
// CHECK18-NEXT: store i64 8, i64* [[TMP138]], align 8 // CHECK18-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8
// CHECK18-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 6 // CHECK18-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to %struct.TT**
// CHECK18-NEXT: store i8* null, i8** [[TMP139]], align 8 // CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP139]], align 8
// CHECK18-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 7 // CHECK18-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8
// CHECK18-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to double** // CHECK18-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to %struct.TT**
// CHECK18-NEXT: store double* [[VLA1]], double** [[TMP141]], align 8 // CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP141]], align 8
// CHECK18-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 7 // CHECK18-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 8
// CHECK18-NEXT: [[TMP143:%.*]] = bitcast i8** [[TMP142]] to double** // CHECK18-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK18-NEXT: store double* [[VLA1]], double** [[TMP143]], align 8 // CHECK18-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9
// CHECK18-NEXT: [[TMP144:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK18-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i64*
// CHECK18-NEXT: store i64 [[TMP97]], i64* [[TMP144]], align 8 // CHECK18-NEXT: store i64 [[TMP93]], i64* [[TMP144]], align 8
// CHECK18-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 7 // CHECK18-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9
// CHECK18-NEXT: store i8* null, i8** [[TMP145]], align 8 // CHECK18-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i64*
// CHECK18-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 8 // CHECK18-NEXT: store i64 [[TMP93]], i64* [[TMP146]], align 8
// CHECK18-NEXT: [[TMP147:%.*]] = bitcast i8** [[TMP146]] to %struct.TT** // CHECK18-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 9
// CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP147]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP147]], align 8
// CHECK18-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 8 // CHECK18-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK18-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to %struct.TT** // CHECK18-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK18-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP149]], align 8
// CHECK18-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK18-NEXT: store i64 16, i64* [[TMP150]], align 8
// CHECK18-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 8
// CHECK18-NEXT: store i8* null, i8** [[TMP151]], align 8
// CHECK18-NEXT: [[TMP152:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 9
// CHECK18-NEXT: [[TMP153:%.*]] = bitcast i8** [[TMP152]] to i64*
// CHECK18-NEXT: store i64 [[TMP93]], i64* [[TMP153]], align 8
// CHECK18-NEXT: [[TMP154:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 9
// CHECK18-NEXT: [[TMP155:%.*]] = bitcast i8** [[TMP154]] to i64*
// CHECK18-NEXT: store i64 [[TMP93]], i64* [[TMP155]], align 8
// CHECK18-NEXT: [[TMP156:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK18-NEXT: store i64 4, i64* [[TMP156]], align 8
// CHECK18-NEXT: [[TMP157:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS31]], i64 0, i64 9
// CHECK18-NEXT: store i8* null, i8** [[TMP157]], align 8
// CHECK18-NEXT: [[TMP158:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS29]], i32 0, i32 0
// CHECK18-NEXT: [[TMP159:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS30]], i32 0, i32 0
// CHECK18-NEXT: [[TMP160:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK18-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK18-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK18-NEXT: [[TMP161:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP158]], i8** [[TMP159]], i64* [[TMP160]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK18-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP148]], i8** [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK18-NEXT: [[TMP162:%.*]] = icmp ne i32 [[TMP161]], 0 // CHECK18-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK18-NEXT: br i1 [[TMP162]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]] // CHECK18-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK18: omp_offload.failed33: // CHECK18: omp_offload.failed33:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT34]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK18: omp_offload.cont34: // CHECK18: omp_offload.cont34:
// CHECK18-NEXT: br label [[OMP_IF_END36:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END36:%.*]]
// CHECK18: omp_if.else35: // CHECK18: omp_if.else35:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i64 [[TMP91]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]], i64 [[TMP93]]) #[[ATTR3]]
// CHECK18-NEXT: br label [[OMP_IF_END36]] // CHECK18-NEXT: br label [[OMP_IF_END36]]
// CHECK18: omp_if.end36: // CHECK18: omp_if.end36:
// CHECK18-NEXT: [[TMP163:%.*]] = load i32, i32* [[A]], align 4 // CHECK18-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4
// CHECK18-NEXT: [[TMP164:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK18-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP164]]) // CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP153]])
// CHECK18-NEXT: ret i32 [[TMP163]] // CHECK18-NEXT: ret i32 [[TMP152]]
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK18-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK18-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 678 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK18-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK18-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK18-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK18-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK18-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK18-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK18-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK18-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK18-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK18-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 15 Lines
// CHECK18-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK18-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK18-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK18-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK18-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK18-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK18-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK18-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK18-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK18-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK18-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK18-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK18-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK18-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK18-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK18-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK18-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK18-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK18-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK18-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK18-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK18-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK18-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK18-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK18-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK18-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK18-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK18-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK18-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK18-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK18-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK18-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK18-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK18-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK18-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK18-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK18-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK18-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK18-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK18-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK18-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK18-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK18-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK18-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK18-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK18-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK18-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK18-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK18-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK18-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK18-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 8
// CHECK18-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK18-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK18-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK18-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK18-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK18-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK18-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8
// CHECK18-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK18-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK18-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK18-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK18-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK18-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK18-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK18-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK18-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK18-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK18-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK18-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK18-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: br label [[OMP_IF_END:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK18: omp_if.else: // CHECK18: omp_if.else:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR3]]
// CHECK18-NEXT: br label [[OMP_IF_END]] // CHECK18-NEXT: br label [[OMP_IF_END]]
// CHECK18: omp_if.end: // CHECK18: omp_if.end:
// CHECK18-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK18-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK18-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK18-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK18-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK18-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK18-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK18-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK18-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK18-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK18-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK18-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK18-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK18-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK18-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK18-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK18-NEXT: ret i32 [[ADD4]] // CHECK18-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK18-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK18-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK18-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK18-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1 // CHECK18-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1
// CHECK18-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK18-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK18-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1 // CHECK18-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1
// CHECK18-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK18-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK18-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK18-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK18-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1 // CHECK18-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1
// CHECK18-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64 // CHECK18-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64
// CHECK18-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK18-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK18-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK18-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK18-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK18-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK18-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK18-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: br label [[OMP_IF_END:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK18: omp_if.else: // CHECK18: omp_if.else:
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK18-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK18-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK18-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK18-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK18-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK18-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK18-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK18-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK18-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: br label [[OMP_IF_END:%.*]] // CHECK18-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK18: omp_if.else: // CHECK18: omp_if.else:
Show All 22 Lines
// CHECK18-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK18-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK18-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK18-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK18-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK18-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK18-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK18-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK18-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK18-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK18-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK18-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] { // CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK18-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK18-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2 // CHECK18-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2
// CHECK18-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK18-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK18-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2 // CHECK18-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2
// CHECK18-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK18-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK18-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK18-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK18-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK18-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK18-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1 // CHECK18-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1
// CHECK18-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK18-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 129 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK18-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK18-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK18-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK18-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK18-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK18-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK18-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK18-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK18-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK18-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK18-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK18-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK18-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK18-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK18-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK18-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK18-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK18-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK18-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4
// CHECK19-NEXT: [[_TMP15:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[_TMP15:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[A_CASTED19:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A_CASTED19:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED20:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__CASTED20:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [10 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [10 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [10 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [10 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [10 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [10 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK19-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[_TMP26:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK19-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK19-NEXT: store i32 0, i32* [[A]], align 4 // CHECK19-NEXT: store i32 0, i32* [[A]], align 4
// CHECK19-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK19-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK19-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK19-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
▲ Show 20 Lines • Show All 162 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[TMP96:%.*]] = mul nuw i32 [[TMP95]], 8 // CHECK19-NEXT: [[TMP96:%.*]] = mul nuw i32 [[TMP95]], 8
// CHECK19-NEXT: [[TMP97:%.*]] = sext i32 [[TMP96]] to i64 // CHECK19-NEXT: [[TMP97:%.*]] = sext i32 [[TMP96]] to i64
// CHECK19-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0 // CHECK19-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK19-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32* // CHECK19-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32*
// CHECK19-NEXT: store i32 [[TMP89]], i32* [[TMP99]], align 4 // CHECK19-NEXT: store i32 [[TMP89]], i32* [[TMP99]], align 4
// CHECK19-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0 // CHECK19-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK19-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i32* // CHECK19-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i32*
// CHECK19-NEXT: store i32 [[TMP89]], i32* [[TMP101]], align 4 // CHECK19-NEXT: store i32 [[TMP89]], i32* [[TMP101]], align 4
// CHECK19-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK19-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 0
// CHECK19-NEXT: store i64 4, i64* [[TMP102]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP102]], align 4
// CHECK19-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 0 // CHECK19-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1
// CHECK19-NEXT: store i8* null, i8** [[TMP103]], align 4 // CHECK19-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]**
// CHECK19-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1 // CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 4
// CHECK19-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to [10 x float]** // CHECK19-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 1
// CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP105]], align 4 // CHECK19-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to [10 x float]**
// CHECK19-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 1 // CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP106]], align 4
// CHECK19-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to [10 x float]** // CHECK19-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 1
// CHECK19-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP107]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP107]], align 4
// CHECK19-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK19-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 2
// CHECK19-NEXT: store i64 40, i64* [[TMP108]], align 4 // CHECK19-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to i32*
// CHECK19-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP109]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP109]], align 4 // CHECK19-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 2
// CHECK19-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 2
// CHECK19-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i32* // CHECK19-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i32*
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP111]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP111]], align 4
// CHECK19-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 2 // CHECK19-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 2
// CHECK19-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i32* // CHECK19-NEXT: store i8* null, i8** [[TMP112]], align 4
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP113]], align 4 // CHECK19-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 3
// CHECK19-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK19-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float**
// CHECK19-NEXT: store i64 4, i64* [[TMP114]], align 4 // CHECK19-NEXT: store float* [[VLA]], float** [[TMP114]], align 4
// CHECK19-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 2 // CHECK19-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 3
// CHECK19-NEXT: store i8* null, i8** [[TMP115]], align 4 // CHECK19-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float**
// CHECK19-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 3 // CHECK19-NEXT: store float* [[VLA]], float** [[TMP116]], align 4
// CHECK19-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to float** // CHECK19-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 3), align 4
// CHECK19-NEXT: store float* [[VLA]], float** [[TMP117]], align 4 // CHECK19-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 3
// CHECK19-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 3 // CHECK19-NEXT: store i8* null, i8** [[TMP117]], align 4
// CHECK19-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to float** // CHECK19-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 4
// CHECK19-NEXT: store float* [[VLA]], float** [[TMP119]], align 4 // CHECK19-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to [5 x [10 x double]]**
// CHECK19-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP119]], align 4
// CHECK19-NEXT: store i64 [[TMP94]], i64* [[TMP120]], align 4 // CHECK19-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 4
// CHECK19-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 3 // CHECK19-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to [5 x [10 x double]]**
// CHECK19-NEXT: store i8* null, i8** [[TMP121]], align 4 // CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP121]], align 4
// CHECK19-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 4 // CHECK19-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 4
// CHECK19-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to [5 x [10 x double]]** // CHECK19-NEXT: store i8* null, i8** [[TMP122]], align 4
// CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP123]], align 4 // CHECK19-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 5
// CHECK19-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 4 // CHECK19-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i32*
// CHECK19-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to [5 x [10 x double]]** // CHECK19-NEXT: store i32 5, i32* [[TMP124]], align 4
// CHECK19-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP125]], align 4 // CHECK19-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 5
// CHECK19-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK19-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32*
// CHECK19-NEXT: store i64 400, i64* [[TMP126]], align 4 // CHECK19-NEXT: store i32 5, i32* [[TMP126]], align 4
// CHECK19-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 4 // CHECK19-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 5
// CHECK19-NEXT: store i8* null, i8** [[TMP127]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP127]], align 4
// CHECK19-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 5 // CHECK19-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 6
// CHECK19-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i32* // CHECK19-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i32*
// CHECK19-NEXT: store i32 5, i32* [[TMP129]], align 4 // CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP129]], align 4
// CHECK19-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 5 // CHECK19-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 6
// CHECK19-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i32* // CHECK19-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i32*
// CHECK19-NEXT: store i32 5, i32* [[TMP131]], align 4 // CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP131]], align 4
// CHECK19-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK19-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 6
// CHECK19-NEXT: store i64 4, i64* [[TMP132]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP132]], align 4
// CHECK19-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 5 // CHECK19-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 7
// CHECK19-NEXT: store i8* null, i8** [[TMP133]], align 4 // CHECK19-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double**
// CHECK19-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 6 // CHECK19-NEXT: store double* [[VLA1]], double** [[TMP134]], align 4
// CHECK19-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i32* // CHECK19-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 7
// CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP135]], align 4 // CHECK19-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to double**
// CHECK19-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 6 // CHECK19-NEXT: store double* [[VLA1]], double** [[TMP136]], align 4
// CHECK19-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i32* // CHECK19-NEXT: store i64 [[TMP97]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 7), align 4
// CHECK19-NEXT: store i32 [[TMP3]], i32* [[TMP137]], align 4 // CHECK19-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 7
// CHECK19-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK19-NEXT: store i8* null, i8** [[TMP137]], align 4
// CHECK19-NEXT: store i64 4, i64* [[TMP138]], align 4 // CHECK19-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 8
// CHECK19-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 6 // CHECK19-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to %struct.TT**
// CHECK19-NEXT: store i8* null, i8** [[TMP139]], align 4 // CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP139]], align 4
// CHECK19-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 7 // CHECK19-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 8
// CHECK19-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to double** // CHECK19-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to %struct.TT**
// CHECK19-NEXT: store double* [[VLA1]], double** [[TMP141]], align 4 // CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP141]], align 4
// CHECK19-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 7 // CHECK19-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 8
// CHECK19-NEXT: [[TMP143:%.*]] = bitcast i8** [[TMP142]] to double** // CHECK19-NEXT: store i8* null, i8** [[TMP142]], align 4
// CHECK19-NEXT: store double* [[VLA1]], double** [[TMP143]], align 4 // CHECK19-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 9
// CHECK19-NEXT: [[TMP144:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK19-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i32*
// CHECK19-NEXT: store i64 [[TMP97]], i64* [[TMP144]], align 4 // CHECK19-NEXT: store i32 [[TMP91]], i32* [[TMP144]], align 4
// CHECK19-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 7 // CHECK19-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 9
// CHECK19-NEXT: store i8* null, i8** [[TMP145]], align 4 // CHECK19-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i32*
// CHECK19-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 8 // CHECK19-NEXT: store i32 [[TMP91]], i32* [[TMP146]], align 4
// CHECK19-NEXT: [[TMP147:%.*]] = bitcast i8** [[TMP146]] to %struct.TT** // CHECK19-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 9
// CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP147]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP147]], align 4
// CHECK19-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 8 // CHECK19-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK19-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to %struct.TT** // CHECK19-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK19-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP149]], align 4
// CHECK19-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK19-NEXT: store i64 12, i64* [[TMP150]], align 4
// CHECK19-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 8
// CHECK19-NEXT: store i8* null, i8** [[TMP151]], align 4
// CHECK19-NEXT: [[TMP152:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 9
// CHECK19-NEXT: [[TMP153:%.*]] = bitcast i8** [[TMP152]] to i32*
// CHECK19-NEXT: store i32 [[TMP91]], i32* [[TMP153]], align 4
// CHECK19-NEXT: [[TMP154:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 9
// CHECK19-NEXT: [[TMP155:%.*]] = bitcast i8** [[TMP154]] to i32*
// CHECK19-NEXT: store i32 [[TMP91]], i32* [[TMP155]], align 4
// CHECK19-NEXT: [[TMP156:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK19-NEXT: store i64 4, i64* [[TMP156]], align 4
// CHECK19-NEXT: [[TMP157:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 9
// CHECK19-NEXT: store i8* null, i8** [[TMP157]], align 4
// CHECK19-NEXT: [[TMP158:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK19-NEXT: [[TMP159:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK19-NEXT: [[TMP160:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK19-NEXT: [[TMP161:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP158]], i8** [[TMP159]], i64* [[TMP160]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK19-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP148]], i8** [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK19-NEXT: [[TMP162:%.*]] = icmp ne i32 [[TMP161]], 0 // CHECK19-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK19-NEXT: br i1 [[TMP162]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]] // CHECK19-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK19: omp_offload.failed27: // CHECK19: omp_offload.failed27:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT28]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK19: omp_offload.cont28: // CHECK19: omp_offload.cont28:
// CHECK19-NEXT: br label [[OMP_IF_END30:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END30:%.*]]
// CHECK19: omp_if.else29: // CHECK19: omp_if.else29:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]]
// CHECK19-NEXT: br label [[OMP_IF_END30]] // CHECK19-NEXT: br label [[OMP_IF_END30]]
// CHECK19: omp_if.end30: // CHECK19: omp_if.end30:
// CHECK19-NEXT: [[TMP163:%.*]] = load i32, i32* [[A]], align 4 // CHECK19-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4
// CHECK19-NEXT: [[TMP164:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP164]]) // CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP153]])
// CHECK19-NEXT: ret i32 [[TMP163]] // CHECK19-NEXT: ret i32 [[TMP152]]
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK19-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK19-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 662 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK19-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK19-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK19-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK19-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK19-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 14 Lines
// CHECK19-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK19-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK19-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK19-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK19-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK19-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK19-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK19-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK19-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK19-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK19-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK19-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK19-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK19-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK19-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK19-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK19-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK19-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK19-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK19-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK19-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK19-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK19-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK19-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK19-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK19-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK19-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK19-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK19-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK19-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK19-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK19-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK19-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK19-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK19-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK19-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK19-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK19-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 4
// CHECK19-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK19-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK19-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK19-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK19-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK19-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK19-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4
// CHECK19-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK19-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK19-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK19-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK19-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK19-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK19-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK19-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK19-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK19-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK19-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: br label [[OMP_IF_END:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK19: omp_if.else: // CHECK19: omp_if.else:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]]
// CHECK19-NEXT: br label [[OMP_IF_END]] // CHECK19-NEXT: br label [[OMP_IF_END]]
// CHECK19: omp_if.end: // CHECK19: omp_if.end:
// CHECK19-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK19-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK19-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK19-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK19-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK19-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK19-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK19-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK19-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK19-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK19-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK19-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK19-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK19-NEXT: ret i32 [[ADD3]] // CHECK19-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK19-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK19-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK19-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1 // CHECK19-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1
// CHECK19-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK19-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK19-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1 // CHECK19-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1
// CHECK19-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK19-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK19-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK19-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK19-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1 // CHECK19-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1
// CHECK19-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64 // CHECK19-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64
// CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK19-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK19-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK19-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK19-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK19-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK19-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: br label [[OMP_IF_END:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK19: omp_if.else: // CHECK19: omp_if.else:
▲ Show 20 Lines • Show All 53 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK19-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK19-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK19-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK19-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK19-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK19-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK19-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK19-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK19-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: br label [[OMP_IF_END:%.*]] // CHECK19-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK19: omp_if.else: // CHECK19: omp_if.else:
Show All 20 Lines
// CHECK19-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK19-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK19-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK19-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK19-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] { // CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 105 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK19-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK19-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK19-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK19-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2 // CHECK19-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2
// CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK19-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK19-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK19-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK19-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK19-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1 // CHECK19-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1
// CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 125 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK19-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK19-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK19-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK19-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK19-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK19-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK19-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 109 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4
// CHECK20-NEXT: [[_TMP15:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[_TMP15:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[A_CASTED19:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A_CASTED19:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTCAPTURE_EXPR__CASTED20:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[DOTCAPTURE_EXPR__CASTED20:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [10 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS23:%.*]] = alloca [10 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [10 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS24:%.*]] = alloca [10 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [10 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS25:%.*]] = alloca [10 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK20-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[_TMP26:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK20-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK20-NEXT: store i32 0, i32* [[A]], align 4 // CHECK20-NEXT: store i32 0, i32* [[A]], align 4
// CHECK20-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK20-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK20-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK20-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
▲ Show 20 Lines • Show All 162 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[TMP96:%.*]] = mul nuw i32 [[TMP95]], 8 // CHECK20-NEXT: [[TMP96:%.*]] = mul nuw i32 [[TMP95]], 8
// CHECK20-NEXT: [[TMP97:%.*]] = sext i32 [[TMP96]] to i64 // CHECK20-NEXT: [[TMP97:%.*]] = sext i32 [[TMP96]] to i64
// CHECK20-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0 // CHECK20-NEXT: [[TMP98:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK20-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32* // CHECK20-NEXT: [[TMP99:%.*]] = bitcast i8** [[TMP98]] to i32*
// CHECK20-NEXT: store i32 [[TMP89]], i32* [[TMP99]], align 4 // CHECK20-NEXT: store i32 [[TMP89]], i32* [[TMP99]], align 4
// CHECK20-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0 // CHECK20-NEXT: [[TMP100:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK20-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i32* // CHECK20-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to i32*
// CHECK20-NEXT: store i32 [[TMP89]], i32* [[TMP101]], align 4 // CHECK20-NEXT: store i32 [[TMP89]], i32* [[TMP101]], align 4
// CHECK20-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK20-NEXT: [[TMP102:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 0
// CHECK20-NEXT: store i64 4, i64* [[TMP102]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP102]], align 4
// CHECK20-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 0 // CHECK20-NEXT: [[TMP103:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1
// CHECK20-NEXT: store i8* null, i8** [[TMP103]], align 4 // CHECK20-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]**
// CHECK20-NEXT: [[TMP104:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 1 // CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 4
// CHECK20-NEXT: [[TMP105:%.*]] = bitcast i8** [[TMP104]] to [10 x float]** // CHECK20-NEXT: [[TMP105:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 1
// CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP105]], align 4 // CHECK20-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to [10 x float]**
// CHECK20-NEXT: [[TMP106:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 1 // CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP106]], align 4
// CHECK20-NEXT: [[TMP107:%.*]] = bitcast i8** [[TMP106]] to [10 x float]** // CHECK20-NEXT: [[TMP107:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 1
// CHECK20-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP107]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP107]], align 4
// CHECK20-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK20-NEXT: [[TMP108:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 2
// CHECK20-NEXT: store i64 40, i64* [[TMP108]], align 4 // CHECK20-NEXT: [[TMP109:%.*]] = bitcast i8** [[TMP108]] to i32*
// CHECK20-NEXT: [[TMP109:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP109]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP109]], align 4 // CHECK20-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 2
// CHECK20-NEXT: [[TMP110:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 2
// CHECK20-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i32* // CHECK20-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to i32*
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP111]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP111]], align 4
// CHECK20-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 2 // CHECK20-NEXT: [[TMP112:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 2
// CHECK20-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to i32* // CHECK20-NEXT: store i8* null, i8** [[TMP112]], align 4
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP113]], align 4 // CHECK20-NEXT: [[TMP113:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 3
// CHECK20-NEXT: [[TMP114:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK20-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float**
// CHECK20-NEXT: store i64 4, i64* [[TMP114]], align 4 // CHECK20-NEXT: store float* [[VLA]], float** [[TMP114]], align 4
// CHECK20-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 2 // CHECK20-NEXT: [[TMP115:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 3
// CHECK20-NEXT: store i8* null, i8** [[TMP115]], align 4 // CHECK20-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float**
// CHECK20-NEXT: [[TMP116:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 3 // CHECK20-NEXT: store float* [[VLA]], float** [[TMP116]], align 4
// CHECK20-NEXT: [[TMP117:%.*]] = bitcast i8** [[TMP116]] to float** // CHECK20-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 3), align 4
// CHECK20-NEXT: store float* [[VLA]], float** [[TMP117]], align 4 // CHECK20-NEXT: [[TMP117:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 3
// CHECK20-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 3 // CHECK20-NEXT: store i8* null, i8** [[TMP117]], align 4
// CHECK20-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to float** // CHECK20-NEXT: [[TMP118:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 4
// CHECK20-NEXT: store float* [[VLA]], float** [[TMP119]], align 4 // CHECK20-NEXT: [[TMP119:%.*]] = bitcast i8** [[TMP118]] to [5 x [10 x double]]**
// CHECK20-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP119]], align 4
// CHECK20-NEXT: store i64 [[TMP94]], i64* [[TMP120]], align 4 // CHECK20-NEXT: [[TMP120:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 4
// CHECK20-NEXT: [[TMP121:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 3 // CHECK20-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to [5 x [10 x double]]**
// CHECK20-NEXT: store i8* null, i8** [[TMP121]], align 4 // CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP121]], align 4
// CHECK20-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 4 // CHECK20-NEXT: [[TMP122:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 4
// CHECK20-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to [5 x [10 x double]]** // CHECK20-NEXT: store i8* null, i8** [[TMP122]], align 4
// CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP123]], align 4 // CHECK20-NEXT: [[TMP123:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 5
// CHECK20-NEXT: [[TMP124:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 4 // CHECK20-NEXT: [[TMP124:%.*]] = bitcast i8** [[TMP123]] to i32*
// CHECK20-NEXT: [[TMP125:%.*]] = bitcast i8** [[TMP124]] to [5 x [10 x double]]** // CHECK20-NEXT: store i32 5, i32* [[TMP124]], align 4
// CHECK20-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP125]], align 4 // CHECK20-NEXT: [[TMP125:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 5
// CHECK20-NEXT: [[TMP126:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK20-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32*
// CHECK20-NEXT: store i64 400, i64* [[TMP126]], align 4 // CHECK20-NEXT: store i32 5, i32* [[TMP126]], align 4
// CHECK20-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 4 // CHECK20-NEXT: [[TMP127:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 5
// CHECK20-NEXT: store i8* null, i8** [[TMP127]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP127]], align 4
// CHECK20-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 5 // CHECK20-NEXT: [[TMP128:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 6
// CHECK20-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i32* // CHECK20-NEXT: [[TMP129:%.*]] = bitcast i8** [[TMP128]] to i32*
// CHECK20-NEXT: store i32 5, i32* [[TMP129]], align 4 // CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP129]], align 4
// CHECK20-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 5 // CHECK20-NEXT: [[TMP130:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 6
// CHECK20-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i32* // CHECK20-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i32*
// CHECK20-NEXT: store i32 5, i32* [[TMP131]], align 4 // CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP131]], align 4
// CHECK20-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK20-NEXT: [[TMP132:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 6
// CHECK20-NEXT: store i64 4, i64* [[TMP132]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP132]], align 4
// CHECK20-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 5 // CHECK20-NEXT: [[TMP133:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 7
// CHECK20-NEXT: store i8* null, i8** [[TMP133]], align 4 // CHECK20-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to double**
// CHECK20-NEXT: [[TMP134:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 6 // CHECK20-NEXT: store double* [[VLA1]], double** [[TMP134]], align 4
// CHECK20-NEXT: [[TMP135:%.*]] = bitcast i8** [[TMP134]] to i32* // CHECK20-NEXT: [[TMP135:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 7
// CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP135]], align 4 // CHECK20-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to double**
// CHECK20-NEXT: [[TMP136:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 6 // CHECK20-NEXT: store double* [[VLA1]], double** [[TMP136]], align 4
// CHECK20-NEXT: [[TMP137:%.*]] = bitcast i8** [[TMP136]] to i32* // CHECK20-NEXT: store i64 [[TMP97]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 7), align 4
// CHECK20-NEXT: store i32 [[TMP3]], i32* [[TMP137]], align 4 // CHECK20-NEXT: [[TMP137:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 7
// CHECK20-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK20-NEXT: store i8* null, i8** [[TMP137]], align 4
// CHECK20-NEXT: store i64 4, i64* [[TMP138]], align 4 // CHECK20-NEXT: [[TMP138:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 8
// CHECK20-NEXT: [[TMP139:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 6 // CHECK20-NEXT: [[TMP139:%.*]] = bitcast i8** [[TMP138]] to %struct.TT**
// CHECK20-NEXT: store i8* null, i8** [[TMP139]], align 4 // CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP139]], align 4
// CHECK20-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 7 // CHECK20-NEXT: [[TMP140:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 8
// CHECK20-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to double** // CHECK20-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to %struct.TT**
// CHECK20-NEXT: store double* [[VLA1]], double** [[TMP141]], align 4 // CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP141]], align 4
// CHECK20-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 7 // CHECK20-NEXT: [[TMP142:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 8
// CHECK20-NEXT: [[TMP143:%.*]] = bitcast i8** [[TMP142]] to double** // CHECK20-NEXT: store i8* null, i8** [[TMP142]], align 4
// CHECK20-NEXT: store double* [[VLA1]], double** [[TMP143]], align 4 // CHECK20-NEXT: [[TMP143:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 9
// CHECK20-NEXT: [[TMP144:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK20-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to i32*
// CHECK20-NEXT: store i64 [[TMP97]], i64* [[TMP144]], align 4 // CHECK20-NEXT: store i32 [[TMP91]], i32* [[TMP144]], align 4
// CHECK20-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 7 // CHECK20-NEXT: [[TMP145:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 9
// CHECK20-NEXT: store i8* null, i8** [[TMP145]], align 4 // CHECK20-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to i32*
// CHECK20-NEXT: [[TMP146:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 8 // CHECK20-NEXT: store i32 [[TMP91]], i32* [[TMP146]], align 4
// CHECK20-NEXT: [[TMP147:%.*]] = bitcast i8** [[TMP146]] to %struct.TT** // CHECK20-NEXT: [[TMP147:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 9
// CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP147]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP147]], align 4
// CHECK20-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 8 // CHECK20-NEXT: [[TMP148:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK20-NEXT: [[TMP149:%.*]] = bitcast i8** [[TMP148]] to %struct.TT** // CHECK20-NEXT: [[TMP149:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK20-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP149]], align 4
// CHECK20-NEXT: [[TMP150:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK20-NEXT: store i64 12, i64* [[TMP150]], align 4
// CHECK20-NEXT: [[TMP151:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 8
// CHECK20-NEXT: store i8* null, i8** [[TMP151]], align 4
// CHECK20-NEXT: [[TMP152:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 9
// CHECK20-NEXT: [[TMP153:%.*]] = bitcast i8** [[TMP152]] to i32*
// CHECK20-NEXT: store i32 [[TMP91]], i32* [[TMP153]], align 4
// CHECK20-NEXT: [[TMP154:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 9
// CHECK20-NEXT: [[TMP155:%.*]] = bitcast i8** [[TMP154]] to i32*
// CHECK20-NEXT: store i32 [[TMP91]], i32* [[TMP155]], align 4
// CHECK20-NEXT: [[TMP156:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK20-NEXT: store i64 4, i64* [[TMP156]], align 4
// CHECK20-NEXT: [[TMP157:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS25]], i32 0, i32 9
// CHECK20-NEXT: store i8* null, i8** [[TMP157]], align 4
// CHECK20-NEXT: [[TMP158:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS23]], i32 0, i32 0
// CHECK20-NEXT: [[TMP159:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS24]], i32 0, i32 0
// CHECK20-NEXT: [[TMP160:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK20-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK20-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK20-NEXT: [[TMP161:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP158]], i8** [[TMP159]], i64* [[TMP160]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK20-NEXT: [[TMP150:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145.region_id, i32 10, i8** [[TMP148]], i8** [[TMP149]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK20-NEXT: [[TMP162:%.*]] = icmp ne i32 [[TMP161]], 0 // CHECK20-NEXT: [[TMP151:%.*]] = icmp ne i32 [[TMP150]], 0
// CHECK20-NEXT: br i1 [[TMP162]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]] // CHECK20-NEXT: br i1 [[TMP151]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK20: omp_offload.failed27: // CHECK20: omp_offload.failed27:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT28]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK20: omp_offload.cont28: // CHECK20: omp_offload.cont28:
// CHECK20-NEXT: br label [[OMP_IF_END30:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END30:%.*]]
// CHECK20: omp_if.else29: // CHECK20: omp_if.else29:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l145(i32 [[TMP89]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]], i32 [[TMP91]]) #[[ATTR3]]
// CHECK20-NEXT: br label [[OMP_IF_END30]] // CHECK20-NEXT: br label [[OMP_IF_END30]]
// CHECK20: omp_if.end30: // CHECK20: omp_if.end30:
// CHECK20-NEXT: [[TMP163:%.*]] = load i32, i32* [[A]], align 4 // CHECK20-NEXT: [[TMP152:%.*]] = load i32, i32* [[A]], align 4
// CHECK20-NEXT: [[TMP164:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: [[TMP153:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP164]]) // CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP153]])
// CHECK20-NEXT: ret i32 [[TMP163]] // CHECK20-NEXT: ret i32 [[TMP152]]
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103 // CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l103
// CHECK20-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK20-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 662 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK20-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK20-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK20-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK20-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK20-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK20-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK20-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK20-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK20-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 14 Lines
// CHECK20-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK20-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK20-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK20-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK20-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK20-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK20-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK20-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK20-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK20-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK20-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK20-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK20-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK20-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK20-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK20-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK20-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK20-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK20-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK20-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK20-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK20-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK20-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK20-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK20-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK20-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK20-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK20-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK20-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK20-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK20-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK20-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK20-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK20-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK20-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK20-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK20-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK20-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK20-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK20-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK20-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK20-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK20-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK20-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK20-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK20-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK20-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK20-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK20-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK20-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK20-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 4
// CHECK20-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK20-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK20-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK20-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK20-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK20-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK20-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4
// CHECK20-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK20-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK20-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK20-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK20-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK20-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK20-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK20-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK20-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK20-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK20-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK20-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK20-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: br label [[OMP_IF_END:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK20: omp_if.else: // CHECK20: omp_if.else:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l218(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR3]]
// CHECK20-NEXT: br label [[OMP_IF_END]] // CHECK20-NEXT: br label [[OMP_IF_END]]
// CHECK20: omp_if.end: // CHECK20: omp_if.end:
// CHECK20-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK20-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK20-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK20-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK20-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK20-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK20-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK20-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK20-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK20-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK20-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK20-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK20-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK20-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK20-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK20-NEXT: ret i32 [[ADD3]] // CHECK20-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK20-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK20-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK20-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1 // CHECK20-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1
// CHECK20-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK20-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK20-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1 // CHECK20-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1
// CHECK20-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK20-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK20-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK20-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK20-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1 // CHECK20-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1
// CHECK20-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64 // CHECK20-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64
// CHECK20-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK20-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK20-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK20-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK20-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK20-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK20-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK20-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l200(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: br label [[OMP_IF_END:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK20: omp_if.else: // CHECK20: omp_if.else:
▲ Show 20 Lines • Show All 53 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK20-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK20-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK20-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK20-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK20-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK20-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK20-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK20-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK20-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK20-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l183(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR3]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: br label [[OMP_IF_END:%.*]] // CHECK20-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK20: omp_if.else: // CHECK20: omp_if.else:
Show All 20 Lines
// CHECK20-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK20-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK20-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK20-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK20-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK20-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK20-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK20-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] { // CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR2]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK20-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK20-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 105 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK20-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK20-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK20-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK20-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2 // CHECK20-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2
// CHECK20-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK20-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK20-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK20-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK20-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK20-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK20-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1 // CHECK20-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1
// CHECK20-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK20-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 125 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK20-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK20-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK20-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK20-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK20-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK20-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK20-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK20-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK20-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] { // CHECK20-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR2]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK20-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK20-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK20-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 3,423 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_collapse_codegen.cpp

Show First 20 Lines • Show All 676 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK9-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK9-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK9-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK9-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK9-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK9-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK9-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK9-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK9-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK9-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK9-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK9-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP54]] // CHECK9-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 188 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 107 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK10-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK10-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK10-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK10-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK10-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK10-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK10-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK10-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK10-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK10-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK10-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK10-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK10-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK10-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK10-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK10-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK10-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP54]] // CHECK10-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 188 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 107 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK11-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK11-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK11-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK11-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK11-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK11-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK11-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK11-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK11-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK11-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP53]] // CHECK11-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 180 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 105 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK12-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK12-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK12-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK12-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK12-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK12-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK12-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK12-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK12-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK12-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK12-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK12-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK12-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK12-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK12-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK12-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK12-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP53]] // CHECK12-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 180 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 96 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_dist_schedule_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 1,511 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS23:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS24:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES25:%.*]] = alloca [4 x i64], align 8 // CHECK9-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_28:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK9-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK9-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK9-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK9-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK9-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 8
// CHECK9-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK9-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK9-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK9-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK9-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK9-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK9-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK9-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP52]])
// CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK9-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9: omp_offload.failed15:
// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK9: omp_offload.cont16:
// CHECK9-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: [[TMP55:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: store i32 [[TMP55]], i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK9-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK9-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP60]]) // CHECK9-NEXT: store i32 [[TMP56]], i32* [[CONV19]], align 4
// CHECK9-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP57:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK9-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK9-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9: omp_offload.failed16: // CHECK9-NEXT: store i32 [[TMP58]], i32* [[CONV20]], align 4
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK9-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK9-NEXT: [[TMP60:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9: omp_offload.cont17: // CHECK9-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP63:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK9-NEXT: store i32 [[TMP63]], i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK9-NEXT: store i64 [[TMP57]], i64* [[TMP62]], align 8
// CHECK9-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32* // CHECK9-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK9-NEXT: store i32 [[TMP64]], i32* [[CONV20]], align 4 // CHECK9-NEXT: store i64 [[TMP57]], i64* [[TMP64]], align 8
// CHECK9-NEXT: [[TMP65:%.*]] = load i64, i64* [[N_CASTED19]], align 8 // CHECK9-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0
// CHECK9-NEXT: [[TMP66:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK9-NEXT: store i8* null, i8** [[TMP65]], align 8
// CHECK9-NEXT: [[CONV21:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK9-NEXT: store i32 [[TMP66]], i32* [[CONV21]], align 4 // CHECK9-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64*
// CHECK9-NEXT: [[TMP67:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP67]], align 8
// CHECK9-NEXT: [[TMP68:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK9-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK9-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK9-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i64* // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP69]], align 8
// CHECK9-NEXT: store i64 [[TMP65]], i64* [[TMP70]], align 8 // CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1
// CHECK9-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK9-NEXT: store i8* null, i8** [[TMP70]], align 8
// CHECK9-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i64* // CHECK9-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP65]], i64* [[TMP72]], align 8 // CHECK9-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32**
// CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP72]], align 8
// CHECK9-NEXT: store i64 4, i64* [[TMP73]], align 8 // CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 0 // CHECK9-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK9-NEXT: store i8* null, i8** [[TMP74]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 8
// CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP60]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i64* // CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP76]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 1 // CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK9-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i64* // CHECK9-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP78]], align 8 // CHECK9-NEXT: store i64 [[TMP59]], i64* [[TMP77]], align 8
// CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 1 // CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK9-NEXT: store i64 8, i64* [[TMP79]], align 8 // CHECK9-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP59]], i64* [[TMP79]], align 8
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3
// CHECK9-NEXT: store i8* null, i8** [[TMP80]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP80]], align 8
// CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 2 // CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32** // CHECK9-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP82]], align 8 // CHECK9-NEXT: [[TMP83:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 2 // CHECK9-NEXT: store i32 [[TMP83]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK9-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 8 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP84]], 0
// CHECK9-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 2 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: store i64 [[TMP68]], i64* [[TMP85]], align 8 // CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK9-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 2 // CHECK9-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP86]], align 8 // CHECK9-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 3 // CHECK9-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP85]], 1
// CHECK9-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK9-NEXT: [[TMP86:%.*]] = zext i32 [[ADD30]] to i64
// CHECK9-NEXT: store i64 [[TMP67]], i64* [[TMP88]], align 8 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP86]])
// CHECK9-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 3 // CHECK9-NEXT: [[TMP87:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP81]], i8** [[TMP82]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK9-NEXT: [[TMP88:%.*]] = icmp ne i32 [[TMP87]], 0
// CHECK9-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK9-NEXT: br i1 [[TMP88]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK9-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 3 // CHECK9: omp_offload.failed31:
// CHECK9-NEXT: store i64 4, i64* [[TMP91]], align 8 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i64 [[TMP57]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP59]]) #[[ATTR3]]
// CHECK9-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 3 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK9-NEXT: store i8* null, i8** [[TMP92]], align 8 // CHECK9: omp_offload.cont32:
// CHECK9-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK9-NEXT: [[TMP89:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP89]])
// CHECK9-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0
// CHECK9-NEXT: [[TMP96:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP96]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP97]], 0
// CHECK9-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK9-NEXT: [[SUB31:%.*]] = sub nsw i32 [[DIV30]], 1
// CHECK9-NEXT: store i32 [[SUB31]], i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK9-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK9-NEXT: [[ADD32:%.*]] = add nsw i32 [[TMP98]], 1
// CHECK9-NEXT: [[TMP99:%.*]] = zext i32 [[ADD32]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP99]])
// CHECK9-NEXT: [[TMP100:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP93]], i8** [[TMP94]], i64* [[TMP95]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP101:%.*]] = icmp ne i32 [[TMP100]], 0
// CHECK9-NEXT: br i1 [[TMP101]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK9: omp_offload.failed33:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i64 [[TMP65]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP67]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK9: omp_offload.cont34:
// CHECK9-NEXT: [[TMP102:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP102]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP103:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP90:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP103]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP90]])
// CHECK9-NEXT: [[TMP104:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP91:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP104]] // CHECK9-NEXT: ret i32 [[TMP91]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 243 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4 // CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK9-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 128 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK9: omp_offload.failed5: // CHECK9: omp_offload.failed5:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK9: omp_offload.cont6: // CHECK9: omp_offload.cont6:
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK9-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK9-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK9: omp_offload.failed11: // CHECK9: omp_offload.failed11:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK9: omp_offload.cont12: // CHECK9: omp_offload.cont12:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 87 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK10-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK10-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS23:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS24:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES25:%.*]] = alloca [4 x i64], align 8 // CHECK10-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_28:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK10-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK10-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK10-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK10-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK10-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK10-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 8
// CHECK10-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK10-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK10-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK10-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK10-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK10-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK10-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK10-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK10-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK10-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK10-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK10-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK10-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK10-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK10-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP52]])
// CHECK10-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK10-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10: omp_offload.failed15:
// CHECK10-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK10-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK10: omp_offload.cont16:
// CHECK10-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: [[TMP55:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: store i32 [[TMP55]], i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK10-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK10-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK10-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP60]]) // CHECK10-NEXT: store i32 [[TMP56]], i32* [[CONV19]], align 4
// CHECK10-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP57:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK10-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK10-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK10-NEXT: [[CONV20:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10: omp_offload.failed16: // CHECK10-NEXT: store i32 [[TMP58]], i32* [[CONV20]], align 4
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK10-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK10-NEXT: [[TMP60:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10: omp_offload.cont17: // CHECK10-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP63:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK10-NEXT: store i32 [[TMP63]], i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK10-NEXT: store i64 [[TMP57]], i64* [[TMP62]], align 8
// CHECK10-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK10-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32* // CHECK10-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK10-NEXT: store i32 [[TMP64]], i32* [[CONV20]], align 4 // CHECK10-NEXT: store i64 [[TMP57]], i64* [[TMP64]], align 8
// CHECK10-NEXT: [[TMP65:%.*]] = load i64, i64* [[N_CASTED19]], align 8 // CHECK10-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0
// CHECK10-NEXT: [[TMP66:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK10-NEXT: store i8* null, i8** [[TMP65]], align 8
// CHECK10-NEXT: [[CONV21:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK10-NEXT: store i32 [[TMP66]], i32* [[CONV21]], align 4 // CHECK10-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64*
// CHECK10-NEXT: [[TMP67:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP67]], align 8
// CHECK10-NEXT: [[TMP68:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK10-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK10-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK10-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i64* // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP69]], align 8
// CHECK10-NEXT: store i64 [[TMP65]], i64* [[TMP70]], align 8 // CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1
// CHECK10-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK10-NEXT: store i8* null, i8** [[TMP70]], align 8
// CHECK10-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i64* // CHECK10-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK10-NEXT: store i64 [[TMP65]], i64* [[TMP72]], align 8 // CHECK10-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32**
// CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP72]], align 8
// CHECK10-NEXT: store i64 4, i64* [[TMP73]], align 8 // CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 0 // CHECK10-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK10-NEXT: store i8* null, i8** [[TMP74]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 8
// CHECK10-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP60]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i64* // CHECK10-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP76]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK10-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 1 // CHECK10-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK10-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i64* // CHECK10-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP78]], align 8 // CHECK10-NEXT: store i64 [[TMP59]], i64* [[TMP77]], align 8
// CHECK10-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 1 // CHECK10-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK10-NEXT: store i64 8, i64* [[TMP79]], align 8 // CHECK10-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP59]], i64* [[TMP79]], align 8
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3
// CHECK10-NEXT: store i8* null, i8** [[TMP80]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP80]], align 8
// CHECK10-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 2 // CHECK10-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32** // CHECK10-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP82]], align 8 // CHECK10-NEXT: [[TMP83:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 2 // CHECK10-NEXT: store i32 [[TMP83]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK10-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 8 // CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP84]], 0
// CHECK10-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 2 // CHECK10-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK10-NEXT: store i64 [[TMP68]], i64* [[TMP85]], align 8 // CHECK10-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK10-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 2 // CHECK10-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP86]], align 8 // CHECK10-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 3 // CHECK10-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP85]], 1
// CHECK10-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK10-NEXT: [[TMP86:%.*]] = zext i32 [[ADD30]] to i64
// CHECK10-NEXT: store i64 [[TMP67]], i64* [[TMP88]], align 8 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP86]])
// CHECK10-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 3 // CHECK10-NEXT: [[TMP87:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP81]], i8** [[TMP82]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK10-NEXT: [[TMP88:%.*]] = icmp ne i32 [[TMP87]], 0
// CHECK10-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK10-NEXT: br i1 [[TMP88]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK10-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 3 // CHECK10: omp_offload.failed31:
// CHECK10-NEXT: store i64 4, i64* [[TMP91]], align 8 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i64 [[TMP57]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP59]]) #[[ATTR3]]
// CHECK10-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 3 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK10-NEXT: store i8* null, i8** [[TMP92]], align 8 // CHECK10: omp_offload.cont32:
// CHECK10-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK10-NEXT: [[TMP89:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP89]])
// CHECK10-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0
// CHECK10-NEXT: [[TMP96:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP96]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP97]], 0
// CHECK10-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK10-NEXT: [[SUB31:%.*]] = sub nsw i32 [[DIV30]], 1
// CHECK10-NEXT: store i32 [[SUB31]], i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK10-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK10-NEXT: [[ADD32:%.*]] = add nsw i32 [[TMP98]], 1
// CHECK10-NEXT: [[TMP99:%.*]] = zext i32 [[ADD32]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP99]])
// CHECK10-NEXT: [[TMP100:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP93]], i8** [[TMP94]], i64* [[TMP95]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP101:%.*]] = icmp ne i32 [[TMP100]], 0
// CHECK10-NEXT: br i1 [[TMP101]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK10: omp_offload.failed33:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i64 [[TMP65]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP67]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK10: omp_offload.cont34:
// CHECK10-NEXT: [[TMP102:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP102]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP103:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP90:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP103]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP90]])
// CHECK10-NEXT: [[TMP104:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP91:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP104]] // CHECK10-NEXT: ret i32 [[TMP91]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 243 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4 // CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK10-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 128 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK10: omp_offload.failed5: // CHECK10: omp_offload.failed5:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK10: omp_offload.cont6: // CHECK10: omp_offload.cont6:
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP22]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK10-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK10-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK10-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK10: omp_offload.failed11: // CHECK10: omp_offload.failed11:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK10: omp_offload.cont12: // CHECK10: omp_offload.cont12:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 87 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK11-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_16:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4 // CHECK11-NEXT: [[_TMP21:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP23:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK11-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK11-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK11-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK11-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK11-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK11-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK11-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK11-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK11-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK11-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP53]])
// CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK11-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11: omp_offload.failed14:
// CHECK11-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK11-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK11: omp_offload.cont15:
// CHECK11-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: store i32 [[TMP56]], i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK11-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK11-NEXT: store i32 [[TMP57]], i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP61]]) // CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK11-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK11-NEXT: store i32 [[TMP59]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11: omp_offload.failed15: // CHECK11-NEXT: [[TMP61:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK11-NEXT: [[TMP62:%.*]] = sext i32 [[TMP61]] to i64
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK11-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK11: omp_offload.cont16: // CHECK11-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32*
// CHECK11-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: store i32 [[TMP58]], i32* [[TMP64]], align 4
// CHECK11-NEXT: store i32 [[TMP64]], i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK11-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK11-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[N_CASTED18]], align 4 // CHECK11-NEXT: store i32 [[TMP58]], i32* [[TMP66]], align 4
// CHECK11-NEXT: [[TMP66:%.*]] = load i32, i32* [[N_CASTED18]], align 4 // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0
// CHECK11-NEXT: [[TMP67:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP67]], align 4
// CHECK11-NEXT: store i32 [[TMP67]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1
// CHECK11-NEXT: [[TMP68:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK11-NEXT: [[TMP69:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP69]], align 4
// CHECK11-NEXT: [[TMP70:%.*]] = sext i32 [[TMP69]] to i64 // CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1
// CHECK11-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK11-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK11-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32* // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP71]], align 4
// CHECK11-NEXT: store i32 [[TMP66]], i32* [[TMP72]], align 4 // CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1
// CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK11-NEXT: store i8* null, i8** [[TMP72]], align 4
// CHECK11-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32* // CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP66]], i32* [[TMP74]], align 4 // CHECK11-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP75]], align 4 // CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2
// CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0 // CHECK11-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK11-NEXT: store i8* null, i8** [[TMP76]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 4
// CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32* // CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP78]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP77]], align 4
// CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1 // CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3
// CHECK11-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK11-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP80]], align 4 // CHECK11-NEXT: store i32 [[TMP60]], i32* [[TMP79]], align 4
// CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 1 // CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3
// CHECK11-NEXT: store i64 4, i64* [[TMP81]], align 4 // CHECK11-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32*
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP60]], i32* [[TMP81]], align 4
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP82]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP82]], align 4
// CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2 // CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK11-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK11-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 4 // CHECK11-NEXT: [[TMP85:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[TMP85]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32** // CHECK11-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP86]], align 4 // CHECK11-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP86]], 0
// CHECK11-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 2 // CHECK11-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK11-NEXT: store i64 [[TMP70]], i64* [[TMP87]], align 4 // CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[DIV25]], 1
// CHECK11-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[SUB26]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP88]], align 4 // CHECK11-NEXT: [[TMP87:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3 // CHECK11-NEXT: [[ADD27:%.*]] = add nsw i32 [[TMP87]], 1
// CHECK11-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK11-NEXT: [[TMP88:%.*]] = zext i32 [[ADD27]] to i64
// CHECK11-NEXT: store i32 [[TMP68]], i32* [[TMP90]], align 4 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP88]])
// CHECK11-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 3 // CHECK11-NEXT: [[TMP89:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP83]], i8** [[TMP84]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32* // CHECK11-NEXT: [[TMP90:%.*]] = icmp ne i32 [[TMP89]], 0
// CHECK11-NEXT: store i32 [[TMP68]], i32* [[TMP92]], align 4 // CHECK11-NEXT: br i1 [[TMP90]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
// CHECK11-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 3 // CHECK11: omp_offload.failed28:
// CHECK11-NEXT: store i64 4, i64* [[TMP93]], align 4 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i32 [[TMP58]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP60]]) #[[ATTR3]]
// CHECK11-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT29]]
// CHECK11-NEXT: store i8* null, i8** [[TMP94]], align 4 // CHECK11: omp_offload.cont29:
// CHECK11-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK11-NEXT: [[TMP91:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP91]])
// CHECK11-NEXT: [[TMP97:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
// CHECK11-NEXT: [[TMP98:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP98]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP99]], 0
// CHECK11-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK11-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK11-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK11-NEXT: [[TMP100:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK11-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP100]], 1
// CHECK11-NEXT: [[TMP101:%.*]] = zext i32 [[ADD29]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP101]])
// CHECK11-NEXT: [[TMP102:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP95]], i8** [[TMP96]], i64* [[TMP97]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP103:%.*]] = icmp ne i32 [[TMP102]], 0
// CHECK11-NEXT: br i1 [[TMP103]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK11: omp_offload.failed30:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i32 [[TMP66]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP68]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK11: omp_offload.cont31:
// CHECK11-NEXT: [[TMP104:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP104]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP105:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP92:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP105]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP92]])
// CHECK11-NEXT: [[TMP106:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP93:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP106]] // CHECK11-NEXT: ret i32 [[TMP93]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 231 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 125 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK11: omp_offload.failed5: // CHECK11: omp_offload.failed5:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK11: omp_offload.cont6: // CHECK11: omp_offload.cont6:
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK11-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK11-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK11-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK11: omp_offload.failed11: // CHECK11: omp_offload.failed11:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK11: omp_offload.cont12: // CHECK11: omp_offload.cont12:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK12-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK12-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_16:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4 // CHECK12-NEXT: [[_TMP21:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP23:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK12-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK12-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK12-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK12-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK12-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK12-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK12-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK12-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK12-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK12-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK12-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK12-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK12-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK12-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK12-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK12-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP53]])
// CHECK12-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK12-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12: omp_offload.failed14:
// CHECK12-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK12-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK12: omp_offload.cont15:
// CHECK12-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: store i32 [[TMP56]], i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK12-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK12-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK12-NEXT: store i32 [[TMP57]], i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP61]]) // CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK12-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK12-NEXT: store i32 [[TMP59]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12: omp_offload.failed15: // CHECK12-NEXT: [[TMP61:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK12-NEXT: [[TMP62:%.*]] = sext i32 [[TMP61]] to i64
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK12-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK12: omp_offload.cont16: // CHECK12-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32*
// CHECK12-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: store i32 [[TMP58]], i32* [[TMP64]], align 4
// CHECK12-NEXT: store i32 [[TMP64]], i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK12-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK12-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[N_CASTED18]], align 4 // CHECK12-NEXT: store i32 [[TMP58]], i32* [[TMP66]], align 4
// CHECK12-NEXT: [[TMP66:%.*]] = load i32, i32* [[N_CASTED18]], align 4 // CHECK12-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0
// CHECK12-NEXT: [[TMP67:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP67]], align 4
// CHECK12-NEXT: store i32 [[TMP67]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1
// CHECK12-NEXT: [[TMP68:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK12-NEXT: [[TMP69:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP69]], align 4
// CHECK12-NEXT: [[TMP70:%.*]] = sext i32 [[TMP69]] to i64 // CHECK12-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1
// CHECK12-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK12-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK12-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32* // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP71]], align 4
// CHECK12-NEXT: store i32 [[TMP66]], i32* [[TMP72]], align 4 // CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1
// CHECK12-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK12-NEXT: store i8* null, i8** [[TMP72]], align 4
// CHECK12-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32* // CHECK12-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP66]], i32* [[TMP74]], align 4 // CHECK12-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP75]], align 4 // CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2
// CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0 // CHECK12-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK12-NEXT: store i8* null, i8** [[TMP76]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 4
// CHECK12-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32* // CHECK12-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP78]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP77]], align 4
// CHECK12-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1 // CHECK12-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3
// CHECK12-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK12-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP80]], align 4 // CHECK12-NEXT: store i32 [[TMP60]], i32* [[TMP79]], align 4
// CHECK12-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 1 // CHECK12-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3
// CHECK12-NEXT: store i64 4, i64* [[TMP81]], align 4 // CHECK12-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32*
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP60]], i32* [[TMP81]], align 4
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP82]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP82]], align 4
// CHECK12-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2 // CHECK12-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK12-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK12-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 4 // CHECK12-NEXT: [[TMP85:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[TMP85]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32** // CHECK12-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP86]], align 4 // CHECK12-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP86]], 0
// CHECK12-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 2 // CHECK12-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK12-NEXT: store i64 [[TMP70]], i64* [[TMP87]], align 4 // CHECK12-NEXT: [[SUB26:%.*]] = sub nsw i32 [[DIV25]], 1
// CHECK12-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[SUB26]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP88]], align 4 // CHECK12-NEXT: [[TMP87:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3 // CHECK12-NEXT: [[ADD27:%.*]] = add nsw i32 [[TMP87]], 1
// CHECK12-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK12-NEXT: [[TMP88:%.*]] = zext i32 [[ADD27]] to i64
// CHECK12-NEXT: store i32 [[TMP68]], i32* [[TMP90]], align 4 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP88]])
// CHECK12-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 3 // CHECK12-NEXT: [[TMP89:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP83]], i8** [[TMP84]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32* // CHECK12-NEXT: [[TMP90:%.*]] = icmp ne i32 [[TMP89]], 0
// CHECK12-NEXT: store i32 [[TMP68]], i32* [[TMP92]], align 4 // CHECK12-NEXT: br i1 [[TMP90]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
// CHECK12-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 3 // CHECK12: omp_offload.failed28:
// CHECK12-NEXT: store i64 4, i64* [[TMP93]], align 4 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i32 [[TMP58]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP60]]) #[[ATTR3]]
// CHECK12-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT29]]
// CHECK12-NEXT: store i8* null, i8** [[TMP94]], align 4 // CHECK12: omp_offload.cont29:
// CHECK12-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK12-NEXT: [[TMP91:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP91]])
// CHECK12-NEXT: [[TMP97:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
// CHECK12-NEXT: [[TMP98:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP98]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP99]], 0
// CHECK12-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK12-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK12-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK12-NEXT: [[TMP100:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK12-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP100]], 1
// CHECK12-NEXT: [[TMP101:%.*]] = zext i32 [[ADD29]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP101]])
// CHECK12-NEXT: [[TMP102:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP95]], i8** [[TMP96]], i64* [[TMP97]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP103:%.*]] = icmp ne i32 [[TMP102]], 0
// CHECK12-NEXT: br i1 [[TMP103]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK12: omp_offload.failed30:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i32 [[TMP66]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP68]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK12: omp_offload.cont31:
// CHECK12-NEXT: [[TMP104:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP104]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP105:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP92:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP105]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP92]])
// CHECK12-NEXT: [[TMP106:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP93:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP106]] // CHECK12-NEXT: ret i32 [[TMP93]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 231 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 125 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK12: omp_offload.failed5: // CHECK12: omp_offload.failed5:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK12: omp_offload.cont6: // CHECK12: omp_offload.cont6:
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK12-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK12-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK12-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK12: omp_offload.failed11: // CHECK12: omp_offload.failed11:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK12: omp_offload.cont12: // CHECK12: omp_offload.cont12:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 80 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_parallel_for_collapse_codegen.cpp

Show First 20 Lines • Show All 973 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK9-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK9-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK9-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK9-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK9-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK9-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK9-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK9-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK9-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK9-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK9-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK9-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP54]] // CHECK9-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 318 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 182 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK10-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK10-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK10-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK10-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK10-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK10-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK10-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK10-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK10-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK10-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK10-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK10-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK10-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK10-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK10-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK10-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK10-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP54]] // CHECK10-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 318 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 182 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK11-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK11-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK11-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK11-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK11-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK11-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK11-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK11-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK11-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK11-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP53]] // CHECK11-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 310 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 176 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK12-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK12-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK12-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK12-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK12-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK12-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK12-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK12-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK12-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK12-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK12-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK12-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK12-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK12-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK12-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK12-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK12-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP53]] // CHECK12-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 310 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 167 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_parallel_for_dist_schedule_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 2,388 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS23:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS24:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES25:%.*]] = alloca [4 x i64], align 8 // CHECK9-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_28:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: store i32 10, i32* [[M]], align 4 // CHECK9-NEXT: store i32 10, i32* [[M]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK9-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK9-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK9-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK9-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK9-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 8
// CHECK9-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK9-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK9-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK9-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK9-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK9-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK9-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK9-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP52]])
// CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK9-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9: omp_offload.failed15:
// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK9: omp_offload.cont16:
// CHECK9-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: [[TMP55:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: store i32 [[TMP55]], i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK9-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK9-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP60]]) // CHECK9-NEXT: store i32 [[TMP56]], i32* [[CONV19]], align 4
// CHECK9-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP57:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK9-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK9-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9: omp_offload.failed16: // CHECK9-NEXT: store i32 [[TMP58]], i32* [[CONV20]], align 4
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK9-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK9-NEXT: [[TMP60:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9: omp_offload.cont17: // CHECK9-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP63:%.*]] = load i32, i32* [[M]], align 4 // CHECK9-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK9-NEXT: store i32 [[TMP63]], i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK9-NEXT: store i64 [[TMP57]], i64* [[TMP62]], align 8
// CHECK9-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32* // CHECK9-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK9-NEXT: store i32 [[TMP64]], i32* [[CONV20]], align 4 // CHECK9-NEXT: store i64 [[TMP57]], i64* [[TMP64]], align 8
// CHECK9-NEXT: [[TMP65:%.*]] = load i64, i64* [[N_CASTED19]], align 8 // CHECK9-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0
// CHECK9-NEXT: [[TMP66:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK9-NEXT: store i8* null, i8** [[TMP65]], align 8
// CHECK9-NEXT: [[CONV21:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK9-NEXT: store i32 [[TMP66]], i32* [[CONV21]], align 4 // CHECK9-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64*
// CHECK9-NEXT: [[TMP67:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP67]], align 8
// CHECK9-NEXT: [[TMP68:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK9-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK9-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK9-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i64* // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP69]], align 8
// CHECK9-NEXT: store i64 [[TMP65]], i64* [[TMP70]], align 8 // CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1
// CHECK9-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK9-NEXT: store i8* null, i8** [[TMP70]], align 8
// CHECK9-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i64* // CHECK9-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP65]], i64* [[TMP72]], align 8 // CHECK9-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32**
// CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP72]], align 8
// CHECK9-NEXT: store i64 4, i64* [[TMP73]], align 8 // CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 0 // CHECK9-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK9-NEXT: store i8* null, i8** [[TMP74]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 8
// CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP60]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i64* // CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP76]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 1 // CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK9-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i64* // CHECK9-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP78]], align 8 // CHECK9-NEXT: store i64 [[TMP59]], i64* [[TMP77]], align 8
// CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 1 // CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK9-NEXT: store i64 8, i64* [[TMP79]], align 8 // CHECK9-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP59]], i64* [[TMP79]], align 8
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3
// CHECK9-NEXT: store i8* null, i8** [[TMP80]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP80]], align 8
// CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 2 // CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32** // CHECK9-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP82]], align 8 // CHECK9-NEXT: [[TMP83:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 2 // CHECK9-NEXT: store i32 [[TMP83]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK9-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 8 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP84]], 0
// CHECK9-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 2 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: store i64 [[TMP68]], i64* [[TMP85]], align 8 // CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK9-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 2 // CHECK9-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP86]], align 8 // CHECK9-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 3 // CHECK9-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP85]], 1
// CHECK9-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK9-NEXT: [[TMP86:%.*]] = zext i32 [[ADD30]] to i64
// CHECK9-NEXT: store i64 [[TMP67]], i64* [[TMP88]], align 8 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP86]])
// CHECK9-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 3 // CHECK9-NEXT: [[TMP87:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP81]], i8** [[TMP82]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK9-NEXT: [[TMP88:%.*]] = icmp ne i32 [[TMP87]], 0
// CHECK9-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK9-NEXT: br i1 [[TMP88]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK9-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 3 // CHECK9: omp_offload.failed31:
// CHECK9-NEXT: store i64 4, i64* [[TMP91]], align 8 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP57]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP59]]) #[[ATTR3]]
// CHECK9-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 3 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK9-NEXT: store i8* null, i8** [[TMP92]], align 8 // CHECK9: omp_offload.cont32:
// CHECK9-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK9-NEXT: [[TMP89:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP89]])
// CHECK9-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0
// CHECK9-NEXT: [[TMP96:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP96]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP97]], 0
// CHECK9-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK9-NEXT: [[SUB31:%.*]] = sub nsw i32 [[DIV30]], 1
// CHECK9-NEXT: store i32 [[SUB31]], i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK9-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK9-NEXT: [[ADD32:%.*]] = add nsw i32 [[TMP98]], 1
// CHECK9-NEXT: [[TMP99:%.*]] = zext i32 [[ADD32]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP99]])
// CHECK9-NEXT: [[TMP100:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP93]], i8** [[TMP94]], i64* [[TMP95]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP101:%.*]] = icmp ne i32 [[TMP100]], 0
// CHECK9-NEXT: br i1 [[TMP101]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK9: omp_offload.failed33:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP65]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP67]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK9: omp_offload.cont34:
// CHECK9-NEXT: [[TMP102:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP102]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP103:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP90:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP103]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP90]])
// CHECK9-NEXT: [[TMP104:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP91:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP104]] // CHECK9-NEXT: ret i32 [[TMP91]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 455 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4 // CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK9-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: [[CONV8:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV8:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP20]], i32* [[CONV8]], align 4 // CHECK9-NEXT: store i32 [[TMP20]], i32* [[CONV8]], align 4
// CHECK9-NEXT: [[TMP21:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[CONV1]], align 4 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK9-NEXT: [[CONV9:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV9:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP22]], i32* [[CONV9]], align 4 // CHECK9-NEXT: store i32 [[TMP22]], i32* [[CONV9]], align 4
// CHECK9-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP17]], i64 [[TMP19]], i64 [[TMP21]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP23]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, i32*, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP17]], i64 [[TMP19]], i64 [[TMP21]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP23]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] // CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP25]]
// CHECK9-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 25 Lines
// CHECK9-NEXT: [[TMP35:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP36:%.*]] = load i32, i32* [[TMP35]], align 4 // CHECK9-NEXT: [[TMP36:%.*]] = load i32, i32* [[TMP35]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP36]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP36]])
// CHECK9-NEXT: br label [[OMP_PRECOND_END]] // CHECK9-NEXT: br label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.end: // CHECK9: omp.precond.end:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 124 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK9: omp_offload.failed5: // CHECK9: omp_offload.failed5:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK9: omp_offload.cont6: // CHECK9: omp_offload.cont6:
// CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4
Show All 15 Lines
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK9-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK9-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64*
// CHECK9-NEXT: store i64 [[TMP20]], i64* [[TMP29]], align 8 // CHECK9-NEXT: store i64 [[TMP20]], i64* [[TMP29]], align 8
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP30]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP30]], align 8
// CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK9-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK9-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK9-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK9: omp_offload.failed11: // CHECK9: omp_offload.failed11:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i64 [[TMP20]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i64 [[TMP20]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK9: omp_offload.cont12: // CHECK9: omp_offload.cont12:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 66 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 74 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32*
// CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP1]], i32* [[CONV1]], align 4 // CHECK9-NEXT: store i32 [[TMP1]], i32* [[CONV1]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP2]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP2]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 39 Lines
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4 // CHECK9-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4
// CHECK9-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 19 Lines
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK10-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS23:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS24:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES25:%.*]] = alloca [4 x i64], align 8 // CHECK10-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_28:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: store i32 10, i32* [[M]], align 4 // CHECK10-NEXT: store i32 10, i32* [[M]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK10-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK10-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK10-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK10-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK10-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK10-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 8
// CHECK10-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK10-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK10-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK10-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK10-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK10-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK10-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK10-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK10-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK10-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK10-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK10-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK10-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK10-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK10-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP52]])
// CHECK10-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK10-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10: omp_offload.failed15:
// CHECK10-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK10-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK10: omp_offload.cont16:
// CHECK10-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: [[TMP55:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: store i32 [[TMP55]], i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK10-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK10-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK10-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP60]]) // CHECK10-NEXT: store i32 [[TMP56]], i32* [[CONV19]], align 4
// CHECK10-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP57:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK10-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK10-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK10-NEXT: [[CONV20:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10: omp_offload.failed16: // CHECK10-NEXT: store i32 [[TMP58]], i32* [[CONV20]], align 4
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK10-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK10-NEXT: [[TMP60:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10: omp_offload.cont17: // CHECK10-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP63:%.*]] = load i32, i32* [[M]], align 4 // CHECK10-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK10-NEXT: store i32 [[TMP63]], i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK10-NEXT: store i64 [[TMP57]], i64* [[TMP62]], align 8
// CHECK10-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK10-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32* // CHECK10-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK10-NEXT: store i32 [[TMP64]], i32* [[CONV20]], align 4 // CHECK10-NEXT: store i64 [[TMP57]], i64* [[TMP64]], align 8
// CHECK10-NEXT: [[TMP65:%.*]] = load i64, i64* [[N_CASTED19]], align 8 // CHECK10-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0
// CHECK10-NEXT: [[TMP66:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK10-NEXT: store i8* null, i8** [[TMP65]], align 8
// CHECK10-NEXT: [[CONV21:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK10-NEXT: store i32 [[TMP66]], i32* [[CONV21]], align 4 // CHECK10-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64*
// CHECK10-NEXT: [[TMP67:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP67]], align 8
// CHECK10-NEXT: [[TMP68:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK10-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK10-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK10-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i64* // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP69]], align 8
// CHECK10-NEXT: store i64 [[TMP65]], i64* [[TMP70]], align 8 // CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1
// CHECK10-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK10-NEXT: store i8* null, i8** [[TMP70]], align 8
// CHECK10-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i64* // CHECK10-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK10-NEXT: store i64 [[TMP65]], i64* [[TMP72]], align 8 // CHECK10-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32**
// CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP72]], align 8
// CHECK10-NEXT: store i64 4, i64* [[TMP73]], align 8 // CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 0 // CHECK10-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK10-NEXT: store i8* null, i8** [[TMP74]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 8
// CHECK10-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP60]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i64* // CHECK10-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP76]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK10-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 1 // CHECK10-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK10-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i64* // CHECK10-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP78]], align 8 // CHECK10-NEXT: store i64 [[TMP59]], i64* [[TMP77]], align 8
// CHECK10-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 1 // CHECK10-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK10-NEXT: store i64 8, i64* [[TMP79]], align 8 // CHECK10-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP59]], i64* [[TMP79]], align 8
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3
// CHECK10-NEXT: store i8* null, i8** [[TMP80]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP80]], align 8
// CHECK10-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 2 // CHECK10-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32** // CHECK10-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP82]], align 8 // CHECK10-NEXT: [[TMP83:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 2 // CHECK10-NEXT: store i32 [[TMP83]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK10-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 8 // CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP84]], 0
// CHECK10-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 2 // CHECK10-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK10-NEXT: store i64 [[TMP68]], i64* [[TMP85]], align 8 // CHECK10-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK10-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 2 // CHECK10-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP86]], align 8 // CHECK10-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 3 // CHECK10-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP85]], 1
// CHECK10-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK10-NEXT: [[TMP86:%.*]] = zext i32 [[ADD30]] to i64
// CHECK10-NEXT: store i64 [[TMP67]], i64* [[TMP88]], align 8 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP86]])
// CHECK10-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 3 // CHECK10-NEXT: [[TMP87:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP81]], i8** [[TMP82]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK10-NEXT: [[TMP88:%.*]] = icmp ne i32 [[TMP87]], 0
// CHECK10-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK10-NEXT: br i1 [[TMP88]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK10-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 3 // CHECK10: omp_offload.failed31:
// CHECK10-NEXT: store i64 4, i64* [[TMP91]], align 8 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP57]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP59]]) #[[ATTR3]]
// CHECK10-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 3 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK10-NEXT: store i8* null, i8** [[TMP92]], align 8 // CHECK10: omp_offload.cont32:
// CHECK10-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK10-NEXT: [[TMP89:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP89]])
// CHECK10-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0
// CHECK10-NEXT: [[TMP96:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP96]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP97]], 0
// CHECK10-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK10-NEXT: [[SUB31:%.*]] = sub nsw i32 [[DIV30]], 1
// CHECK10-NEXT: store i32 [[SUB31]], i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK10-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK10-NEXT: [[ADD32:%.*]] = add nsw i32 [[TMP98]], 1
// CHECK10-NEXT: [[TMP99:%.*]] = zext i32 [[ADD32]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP99]])
// CHECK10-NEXT: [[TMP100:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP93]], i8** [[TMP94]], i64* [[TMP95]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP101:%.*]] = icmp ne i32 [[TMP100]], 0
// CHECK10-NEXT: br i1 [[TMP101]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK10: omp_offload.failed33:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP65]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP67]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK10: omp_offload.cont34:
// CHECK10-NEXT: [[TMP102:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP102]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP103:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP90:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP103]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP90]])
// CHECK10-NEXT: [[TMP104:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP91:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP104]] // CHECK10-NEXT: ret i32 [[TMP91]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 455 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4 // CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK10-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP20:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP20:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: [[CONV8:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV8:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP20]], i32* [[CONV8]], align 4 // CHECK10-NEXT: store i32 [[TMP20]], i32* [[CONV8]], align 4
// CHECK10-NEXT: [[TMP21:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP21:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = load i32, i32* [[CONV1]], align 4 // CHECK10-NEXT: [[TMP22:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK10-NEXT: [[CONV9:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV9:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP22]], i32* [[CONV9]], align 4 // CHECK10-NEXT: store i32 [[TMP22]], i32* [[CONV9]], align 4
// CHECK10-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP17]], i64 [[TMP19]], i64 [[TMP21]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP23]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, i32*, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP17]], i64 [[TMP19]], i64 [[TMP21]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP23]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] // CHECK10-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP25]]
// CHECK10-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 25 Lines
// CHECK10-NEXT: [[TMP35:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP35:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP36:%.*]] = load i32, i32* [[TMP35]], align 4 // CHECK10-NEXT: [[TMP36:%.*]] = load i32, i32* [[TMP35]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP36]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP36]])
// CHECK10-NEXT: br label [[OMP_PRECOND_END]] // CHECK10-NEXT: br label [[OMP_PRECOND_END]]
// CHECK10: omp.precond.end: // CHECK10: omp.precond.end:
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 124 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK10: omp_offload.failed5: // CHECK10: omp_offload.failed5:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK10: omp_offload.cont6: // CHECK10: omp_offload.cont6:
// CHECK10-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK10-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4
Show All 15 Lines
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK10-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK10-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64*
// CHECK10-NEXT: store i64 [[TMP20]], i64* [[TMP29]], align 8 // CHECK10-NEXT: store i64 [[TMP20]], i64* [[TMP29]], align 8
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1 // CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP30]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP30]], align 8
// CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK10-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK10-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK10-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK10: omp_offload.failed11: // CHECK10: omp_offload.failed11:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i64 [[TMP20]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i64 [[TMP20]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK10: omp_offload.cont12: // CHECK10: omp_offload.cont12:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 66 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 74 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32*
// CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP1]], i32* [[CONV1]], align 4 // CHECK10-NEXT: store i32 [[TMP1]], i32* [[CONV1]], align 4
// CHECK10-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP2]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP2]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 39 Lines
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4 // CHECK10-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4
// CHECK10-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 19 Lines
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK11-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_16:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4 // CHECK11-NEXT: [[_TMP21:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP23:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: store i32 10, i32* [[M]], align 4 // CHECK11-NEXT: store i32 10, i32* [[M]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK11-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK11-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK11-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK11-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK11-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK11-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK11-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK11-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK11-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK11-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP53]])
// CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK11-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11: omp_offload.failed14:
// CHECK11-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK11-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK11: omp_offload.cont15:
// CHECK11-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: [[TMP56:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: store i32 [[TMP56]], i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK11-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK11-NEXT: store i32 [[TMP57]], i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP61]]) // CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK11-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK11-NEXT: store i32 [[TMP59]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11: omp_offload.failed15: // CHECK11-NEXT: [[TMP61:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK11-NEXT: [[TMP62:%.*]] = sext i32 [[TMP61]] to i64
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK11-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK11: omp_offload.cont16: // CHECK11-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32*
// CHECK11-NEXT: [[TMP64:%.*]] = load i32, i32* [[M]], align 4 // CHECK11-NEXT: store i32 [[TMP58]], i32* [[TMP64]], align 4
// CHECK11-NEXT: store i32 [[TMP64]], i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK11-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK11-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[N_CASTED18]], align 4 // CHECK11-NEXT: store i32 [[TMP58]], i32* [[TMP66]], align 4
// CHECK11-NEXT: [[TMP66:%.*]] = load i32, i32* [[N_CASTED18]], align 4 // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0
// CHECK11-NEXT: [[TMP67:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP67]], align 4
// CHECK11-NEXT: store i32 [[TMP67]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1
// CHECK11-NEXT: [[TMP68:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK11-NEXT: [[TMP69:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP69]], align 4
// CHECK11-NEXT: [[TMP70:%.*]] = sext i32 [[TMP69]] to i64 // CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1
// CHECK11-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK11-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK11-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32* // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP71]], align 4
// CHECK11-NEXT: store i32 [[TMP66]], i32* [[TMP72]], align 4 // CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1
// CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK11-NEXT: store i8* null, i8** [[TMP72]], align 4
// CHECK11-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32* // CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP66]], i32* [[TMP74]], align 4 // CHECK11-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP75]], align 4 // CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2
// CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0 // CHECK11-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK11-NEXT: store i8* null, i8** [[TMP76]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 4
// CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32* // CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP78]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP77]], align 4
// CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1 // CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3
// CHECK11-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK11-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP80]], align 4 // CHECK11-NEXT: store i32 [[TMP60]], i32* [[TMP79]], align 4
// CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 1 // CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3
// CHECK11-NEXT: store i64 4, i64* [[TMP81]], align 4 // CHECK11-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32*
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP60]], i32* [[TMP81]], align 4
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP82]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP82]], align 4
// CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2 // CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK11-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK11-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 4 // CHECK11-NEXT: [[TMP85:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[TMP85]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32** // CHECK11-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP86]], align 4 // CHECK11-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP86]], 0
// CHECK11-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 2 // CHECK11-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK11-NEXT: store i64 [[TMP70]], i64* [[TMP87]], align 4 // CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[DIV25]], 1
// CHECK11-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[SUB26]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP88]], align 4 // CHECK11-NEXT: [[TMP87:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3 // CHECK11-NEXT: [[ADD27:%.*]] = add nsw i32 [[TMP87]], 1
// CHECK11-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK11-NEXT: [[TMP88:%.*]] = zext i32 [[ADD27]] to i64
// CHECK11-NEXT: store i32 [[TMP68]], i32* [[TMP90]], align 4 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP88]])
// CHECK11-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 3 // CHECK11-NEXT: [[TMP89:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP83]], i8** [[TMP84]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32* // CHECK11-NEXT: [[TMP90:%.*]] = icmp ne i32 [[TMP89]], 0
// CHECK11-NEXT: store i32 [[TMP68]], i32* [[TMP92]], align 4 // CHECK11-NEXT: br i1 [[TMP90]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
// CHECK11-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 3 // CHECK11: omp_offload.failed28:
// CHECK11-NEXT: store i64 4, i64* [[TMP93]], align 4 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP58]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP60]]) #[[ATTR3]]
// CHECK11-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT29]]
// CHECK11-NEXT: store i8* null, i8** [[TMP94]], align 4 // CHECK11: omp_offload.cont29:
// CHECK11-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK11-NEXT: [[TMP91:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP91]])
// CHECK11-NEXT: [[TMP97:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
// CHECK11-NEXT: [[TMP98:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP98]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP99]], 0
// CHECK11-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK11-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK11-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK11-NEXT: [[TMP100:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK11-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP100]], 1
// CHECK11-NEXT: [[TMP101:%.*]] = zext i32 [[ADD29]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP101]])
// CHECK11-NEXT: [[TMP102:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP95]], i8** [[TMP96]], i64* [[TMP97]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP103:%.*]] = icmp ne i32 [[TMP102]], 0
// CHECK11-NEXT: br i1 [[TMP103]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK11: omp_offload.failed30:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP66]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP68]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK11: omp_offload.cont31:
// CHECK11-NEXT: [[TMP104:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP104]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP105:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP92:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP105]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP92]])
// CHECK11-NEXT: [[TMP106:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP93:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP106]] // CHECK11-NEXT: ret i32 [[TMP93]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 431 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 64 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP18]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP18]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP20]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP20]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP16]], i32 [[TMP17]], i32 [[TMP19]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP21]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, i32*, i32)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32 [[TMP16]], i32 [[TMP17]], i32 [[TMP19]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP21]])
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP23]] // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
// CHECK11-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 25 Lines
// CHECK11-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK11-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 // CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]])
// CHECK11-NEXT: br label [[OMP_PRECOND_END]] // CHECK11-NEXT: br label [[OMP_PRECOND_END]]
// CHECK11: omp.precond.end: // CHECK11: omp.precond.end:
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 119 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK11: omp_offload.failed5: // CHECK11: omp_offload.failed5:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK11: omp_offload.cont6: // CHECK11: omp_offload.cont6:
// CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4
Show All 14 Lines
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK11-NEXT: store i32 [[TMP20]], i32* [[TMP29]], align 4 // CHECK11-NEXT: store i32 [[TMP20]], i32* [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1 // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK11-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK11-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK11-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK11: omp_offload.failed11: // CHECK11: omp_offload.failed11:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i32 [[TMP20]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i32 [[TMP20]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK11: omp_offload.cont12: // CHECK11: omp_offload.cont12:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]])
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]])
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 69 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP2]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP2]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 35 Lines
// CHECK11-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10 // CHECK11-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]])
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 19 Lines
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 80 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK12-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_16:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4 // CHECK12-NEXT: [[_TMP21:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP23:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: store i32 10, i32* [[M]], align 4 // CHECK12-NEXT: store i32 10, i32* [[M]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK12-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK12-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK12-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK12-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK12-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK12-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK12-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK12-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK12-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK12-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK12-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK12-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK12-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK12-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK12-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK12-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP53]])
// CHECK12-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK12-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12: omp_offload.failed14:
// CHECK12-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK12-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK12: omp_offload.cont15:
// CHECK12-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: [[TMP56:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: store i32 [[TMP56]], i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK12-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK12-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK12-NEXT: store i32 [[TMP57]], i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP61]]) // CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK12-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK12-NEXT: store i32 [[TMP59]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12: omp_offload.failed15: // CHECK12-NEXT: [[TMP61:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK12-NEXT: [[TMP62:%.*]] = sext i32 [[TMP61]] to i64
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK12-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK12: omp_offload.cont16: // CHECK12-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32*
// CHECK12-NEXT: [[TMP64:%.*]] = load i32, i32* [[M]], align 4 // CHECK12-NEXT: store i32 [[TMP58]], i32* [[TMP64]], align 4
// CHECK12-NEXT: store i32 [[TMP64]], i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK12-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK12-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[N_CASTED18]], align 4 // CHECK12-NEXT: store i32 [[TMP58]], i32* [[TMP66]], align 4
// CHECK12-NEXT: [[TMP66:%.*]] = load i32, i32* [[N_CASTED18]], align 4 // CHECK12-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0
// CHECK12-NEXT: [[TMP67:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP67]], align 4
// CHECK12-NEXT: store i32 [[TMP67]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1
// CHECK12-NEXT: [[TMP68:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK12-NEXT: [[TMP69:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP69]], align 4
// CHECK12-NEXT: [[TMP70:%.*]] = sext i32 [[TMP69]] to i64 // CHECK12-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1
// CHECK12-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK12-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK12-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32* // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP71]], align 4
// CHECK12-NEXT: store i32 [[TMP66]], i32* [[TMP72]], align 4 // CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1
// CHECK12-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK12-NEXT: store i8* null, i8** [[TMP72]], align 4
// CHECK12-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32* // CHECK12-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP66]], i32* [[TMP74]], align 4 // CHECK12-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP75]], align 4 // CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2
// CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0 // CHECK12-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK12-NEXT: store i8* null, i8** [[TMP76]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 4
// CHECK12-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32* // CHECK12-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP78]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP77]], align 4
// CHECK12-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1 // CHECK12-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3
// CHECK12-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK12-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP80]], align 4 // CHECK12-NEXT: store i32 [[TMP60]], i32* [[TMP79]], align 4
// CHECK12-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 1 // CHECK12-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3
// CHECK12-NEXT: store i64 4, i64* [[TMP81]], align 4 // CHECK12-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32*
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP60]], i32* [[TMP81]], align 4
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP82]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP82]], align 4
// CHECK12-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2 // CHECK12-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK12-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK12-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 4 // CHECK12-NEXT: [[TMP85:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[TMP85]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32** // CHECK12-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP86]], align 4 // CHECK12-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP86]], 0
// CHECK12-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 2 // CHECK12-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK12-NEXT: store i64 [[TMP70]], i64* [[TMP87]], align 4 // CHECK12-NEXT: [[SUB26:%.*]] = sub nsw i32 [[DIV25]], 1
// CHECK12-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[SUB26]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP88]], align 4 // CHECK12-NEXT: [[TMP87:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3 // CHECK12-NEXT: [[ADD27:%.*]] = add nsw i32 [[TMP87]], 1
// CHECK12-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK12-NEXT: [[TMP88:%.*]] = zext i32 [[ADD27]] to i64
// CHECK12-NEXT: store i32 [[TMP68]], i32* [[TMP90]], align 4 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP88]])
// CHECK12-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 3 // CHECK12-NEXT: [[TMP89:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP83]], i8** [[TMP84]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32* // CHECK12-NEXT: [[TMP90:%.*]] = icmp ne i32 [[TMP89]], 0
// CHECK12-NEXT: store i32 [[TMP68]], i32* [[TMP92]], align 4 // CHECK12-NEXT: br i1 [[TMP90]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
// CHECK12-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 3 // CHECK12: omp_offload.failed28:
// CHECK12-NEXT: store i64 4, i64* [[TMP93]], align 4 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP58]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP60]]) #[[ATTR3]]
// CHECK12-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT29]]
// CHECK12-NEXT: store i8* null, i8** [[TMP94]], align 4 // CHECK12: omp_offload.cont29:
// CHECK12-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK12-NEXT: [[TMP91:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP91]])
// CHECK12-NEXT: [[TMP97:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
// CHECK12-NEXT: [[TMP98:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP98]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP99]], 0
// CHECK12-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK12-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK12-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK12-NEXT: [[TMP100:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK12-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP100]], 1
// CHECK12-NEXT: [[TMP101:%.*]] = zext i32 [[ADD29]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP101]])
// CHECK12-NEXT: [[TMP102:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP95]], i8** [[TMP96]], i64* [[TMP97]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP103:%.*]] = icmp ne i32 [[TMP102]], 0
// CHECK12-NEXT: br i1 [[TMP103]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK12: omp_offload.failed30:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP66]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP68]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK12: omp_offload.cont31:
// CHECK12-NEXT: [[TMP104:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP104]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP105:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP92:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP105]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP92]])
// CHECK12-NEXT: [[TMP106:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP93:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP106]] // CHECK12-NEXT: ret i32 [[TMP93]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 431 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 64 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP18]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP18]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP20]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP20]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP16]], i32 [[TMP17]], i32 [[TMP19]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP21]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, i32*, i32)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32 [[TMP16]], i32 [[TMP17]], i32 [[TMP19]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP21]])
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK12-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP23]] // CHECK12-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
// CHECK12-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 25 Lines
// CHECK12-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK12-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 // CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]])
// CHECK12-NEXT: br label [[OMP_PRECOND_END]] // CHECK12-NEXT: br label [[OMP_PRECOND_END]]
// CHECK12: omp.precond.end: // CHECK12: omp.precond.end:
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 119 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK12: omp_offload.failed5: // CHECK12: omp_offload.failed5:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK12: omp_offload.cont6: // CHECK12: omp_offload.cont6:
// CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4
Show All 14 Lines
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK12-NEXT: store i32 [[TMP20]], i32* [[TMP29]], align 4 // CHECK12-NEXT: store i32 [[TMP20]], i32* [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1 // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK12-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK12-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK12-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK12: omp_offload.failed11: // CHECK12: omp_offload.failed11:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i32 [[TMP20]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i32 [[TMP20]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK12: omp_offload.cont12: // CHECK12: omp_offload.cont12:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]])
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]])
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 69 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP2]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP2]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 35 Lines
// CHECK12-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10 // CHECK12-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]])
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 19 Lines
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 73 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_parallel_for_schedule_codegen.cpp

This file has a very large number of changes (5,472 lines). Show File Contents

clang/test/OpenMP/target_teams_distribute_parallel_for_simd_collapse_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 1,317 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK9-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK9-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK9-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK9-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK9-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK9-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK9-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK9-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK9-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK9-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK9-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK9-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP54]] // CHECK9-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 354 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 198 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK10-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK10-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK10-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK10-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK10-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK10-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK10-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK10-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK10-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK10-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK10-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK10-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK10-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK10-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK10-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK10-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK10-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP54]] // CHECK10-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 354 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 198 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK11-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK11-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK11-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK11-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK11-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK11-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK11-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK11-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK11-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK11-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP53]] // CHECK11-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 346 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 192 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK12-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK12-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK12-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK12-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK12-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK12-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK12-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK12-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK12-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK12-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK12-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK12-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK12-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK12-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK12-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK12-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK12-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP53]] // CHECK12-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 346 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 983 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_parallel_for_simd_dist_schedule_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 3,050 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS23:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS24:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES25:%.*]] = alloca [4 x i64], align 8 // CHECK9-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_28:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: store i32 10, i32* [[M]], align 4 // CHECK9-NEXT: store i32 10, i32* [[M]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK9-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK9-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK9-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK9-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK9-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 8
// CHECK9-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK9-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK9-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK9-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK9-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK9-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK9-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK9-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP52]])
// CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK9-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9: omp_offload.failed15:
// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK9: omp_offload.cont16:
// CHECK9-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: [[TMP55:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: store i32 [[TMP55]], i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK9-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK9-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP60]]) // CHECK9-NEXT: store i32 [[TMP56]], i32* [[CONV19]], align 4
// CHECK9-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP57:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK9-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK9-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9: omp_offload.failed16: // CHECK9-NEXT: store i32 [[TMP58]], i32* [[CONV20]], align 4
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK9-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK9-NEXT: [[TMP60:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9: omp_offload.cont17: // CHECK9-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP63:%.*]] = load i32, i32* [[M]], align 4 // CHECK9-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK9-NEXT: store i32 [[TMP63]], i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK9-NEXT: store i64 [[TMP57]], i64* [[TMP62]], align 8
// CHECK9-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32* // CHECK9-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK9-NEXT: store i32 [[TMP64]], i32* [[CONV20]], align 4 // CHECK9-NEXT: store i64 [[TMP57]], i64* [[TMP64]], align 8
// CHECK9-NEXT: [[TMP65:%.*]] = load i64, i64* [[N_CASTED19]], align 8 // CHECK9-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0
// CHECK9-NEXT: [[TMP66:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK9-NEXT: store i8* null, i8** [[TMP65]], align 8
// CHECK9-NEXT: [[CONV21:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK9-NEXT: store i32 [[TMP66]], i32* [[CONV21]], align 4 // CHECK9-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64*
// CHECK9-NEXT: [[TMP67:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP67]], align 8
// CHECK9-NEXT: [[TMP68:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK9-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK9-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK9-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i64* // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP69]], align 8
// CHECK9-NEXT: store i64 [[TMP65]], i64* [[TMP70]], align 8 // CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1
// CHECK9-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK9-NEXT: store i8* null, i8** [[TMP70]], align 8
// CHECK9-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i64* // CHECK9-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP65]], i64* [[TMP72]], align 8 // CHECK9-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32**
// CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP72]], align 8
// CHECK9-NEXT: store i64 4, i64* [[TMP73]], align 8 // CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 0 // CHECK9-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK9-NEXT: store i8* null, i8** [[TMP74]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 8
// CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP60]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i64* // CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP76]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 1 // CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK9-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i64* // CHECK9-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP78]], align 8 // CHECK9-NEXT: store i64 [[TMP59]], i64* [[TMP77]], align 8
// CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 1 // CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK9-NEXT: store i64 8, i64* [[TMP79]], align 8 // CHECK9-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP59]], i64* [[TMP79]], align 8
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3
// CHECK9-NEXT: store i8* null, i8** [[TMP80]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP80]], align 8
// CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 2 // CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32** // CHECK9-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP82]], align 8 // CHECK9-NEXT: [[TMP83:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 2 // CHECK9-NEXT: store i32 [[TMP83]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK9-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 8 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP84]], 0
// CHECK9-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 2 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: store i64 [[TMP68]], i64* [[TMP85]], align 8 // CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK9-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 2 // CHECK9-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP86]], align 8 // CHECK9-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 3 // CHECK9-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP85]], 1
// CHECK9-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK9-NEXT: [[TMP86:%.*]] = zext i32 [[ADD30]] to i64
// CHECK9-NEXT: store i64 [[TMP67]], i64* [[TMP88]], align 8 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP86]])
// CHECK9-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 3 // CHECK9-NEXT: [[TMP87:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP81]], i8** [[TMP82]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK9-NEXT: [[TMP88:%.*]] = icmp ne i32 [[TMP87]], 0
// CHECK9-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK9-NEXT: br i1 [[TMP88]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK9-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 3 // CHECK9: omp_offload.failed31:
// CHECK9-NEXT: store i64 4, i64* [[TMP91]], align 8 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP57]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP59]]) #[[ATTR3]]
// CHECK9-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 3 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK9-NEXT: store i8* null, i8** [[TMP92]], align 8 // CHECK9: omp_offload.cont32:
// CHECK9-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK9-NEXT: [[TMP89:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP89]])
// CHECK9-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0
// CHECK9-NEXT: [[TMP96:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP96]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP97]], 0
// CHECK9-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK9-NEXT: [[SUB31:%.*]] = sub nsw i32 [[DIV30]], 1
// CHECK9-NEXT: store i32 [[SUB31]], i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK9-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK9-NEXT: [[ADD32:%.*]] = add nsw i32 [[TMP98]], 1
// CHECK9-NEXT: [[TMP99:%.*]] = zext i32 [[ADD32]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP99]])
// CHECK9-NEXT: [[TMP100:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP93]], i8** [[TMP94]], i64* [[TMP95]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP101:%.*]] = icmp ne i32 [[TMP100]], 0
// CHECK9-NEXT: br i1 [[TMP101]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK9: omp_offload.failed33:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP65]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP67]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK9: omp_offload.cont34:
// CHECK9-NEXT: [[TMP102:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP102]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP103:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP90:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP103]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP90]])
// CHECK9-NEXT: [[TMP104:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP91:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP104]] // CHECK9-NEXT: ret i32 [[TMP91]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 503 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4 // CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK9-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[CONV8:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV8:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP20]], i32* [[CONV8]], align 4, !llvm.access.group !24 // CHECK9-NEXT: store i32 [[TMP20]], i32* [[CONV8]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP21:%.*]] = load i64, i64* [[N_CASTED]], align 8, !llvm.access.group !24 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, i64* [[N_CASTED]], align 8, !llvm.access.group !24
// CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[CONV1]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[CONV1]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[CONV9:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV9:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP22]], i32* [[CONV9]], align 4, !llvm.access.group !24 // CHECK9-NEXT: store i32 [[TMP22]], i32* [[CONV9]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !24 // CHECK9-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !24
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP17]], i64 [[TMP19]], i64 [[TMP21]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP23]]), !llvm.access.group !24 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, i32*, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP17]], i64 [[TMP19]], i64 [[TMP21]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP23]]), !llvm.access.group !24
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] // CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP25]]
// CHECK9-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK9-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24
Show All 37 Lines
// CHECK9-NEXT: store i32 [[ADD20]], i32* [[I5]], align 4 // CHECK9-NEXT: store i32 [[ADD20]], i32* [[I5]], align 4
// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK9: .omp.final.done: // CHECK9: .omp.final.done:
// CHECK9-NEXT: br label [[OMP_PRECOND_END]] // CHECK9-NEXT: br label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.end: // CHECK9: omp.precond.end:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 136 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK9: omp_offload.failed5: // CHECK9: omp_offload.failed5:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK9: omp_offload.cont6: // CHECK9: omp_offload.cont6:
// CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4
Show All 15 Lines
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK9-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK9-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64*
// CHECK9-NEXT: store i64 [[TMP20]], i64* [[TMP29]], align 8 // CHECK9-NEXT: store i64 [[TMP20]], i64* [[TMP29]], align 8
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP30]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP30]], align 8
// CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK9-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK9-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK9-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK9: omp_offload.failed11: // CHECK9: omp_offload.failed11:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i64 [[TMP20]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i64 [[TMP20]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK9: omp_offload.cont12: // CHECK9: omp_offload.cont12:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !30 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !30
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !30 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !30
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !30 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !30
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
// CHECK9-NEXT: store i32 10, i32* [[I]], align 4 // CHECK9-NEXT: store i32 10, i32* [[I]], align 4
// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK9: .omp.final.done: // CHECK9: .omp.final.done:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !36 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !36
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !36 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !36
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !36 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !36
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
// CHECK9-NEXT: store i32 10, i32* [[I]], align 4 // CHECK9-NEXT: store i32 10, i32* [[I]], align 4
// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK9: .omp.final.done: // CHECK9: .omp.final.done:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 81 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32*
// CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP1]], i32* [[CONV1]], align 4 // CHECK9-NEXT: store i32 [[TMP1]], i32* [[CONV1]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP2]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP2]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 39 Lines
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4, !llvm.access.group !42 // CHECK9-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !42 // CHECK9-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !42
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]), !llvm.access.group !42 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]), !llvm.access.group !42
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42
Show All 26 Lines
// CHECK9-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
// CHECK9-NEXT: store i32 10, i32* [[I]], align 4 // CHECK9-NEXT: store i32 10, i32* [[I]], align 4
// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK9: .omp.final.done: // CHECK9: .omp.final.done:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 91 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK10-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS23:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS24:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES25:%.*]] = alloca [4 x i64], align 8 // CHECK10-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_28:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: store i32 10, i32* [[M]], align 4 // CHECK10-NEXT: store i32 10, i32* [[M]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK10-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK10-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK10-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK10-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK10-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK10-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 8
// CHECK10-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK10-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK10-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK10-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK10-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK10-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK10-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK10-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK10-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK10-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK10-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK10-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK10-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK10-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK10-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP52]])
// CHECK10-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK10-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10: omp_offload.failed15:
// CHECK10-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK10-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK10: omp_offload.cont16:
// CHECK10-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: [[TMP55:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: store i32 [[TMP55]], i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK10-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK10-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK10-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP60]]) // CHECK10-NEXT: store i32 [[TMP56]], i32* [[CONV19]], align 4
// CHECK10-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP57:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK10-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK10-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK10-NEXT: [[CONV20:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10: omp_offload.failed16: // CHECK10-NEXT: store i32 [[TMP58]], i32* [[CONV20]], align 4
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK10-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK10-NEXT: [[TMP60:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10: omp_offload.cont17: // CHECK10-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP63:%.*]] = load i32, i32* [[M]], align 4 // CHECK10-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK10-NEXT: store i32 [[TMP63]], i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK10-NEXT: store i64 [[TMP57]], i64* [[TMP62]], align 8
// CHECK10-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK10-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32* // CHECK10-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK10-NEXT: store i32 [[TMP64]], i32* [[CONV20]], align 4 // CHECK10-NEXT: store i64 [[TMP57]], i64* [[TMP64]], align 8
// CHECK10-NEXT: [[TMP65:%.*]] = load i64, i64* [[N_CASTED19]], align 8 // CHECK10-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0
// CHECK10-NEXT: [[TMP66:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK10-NEXT: store i8* null, i8** [[TMP65]], align 8
// CHECK10-NEXT: [[CONV21:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK10-NEXT: store i32 [[TMP66]], i32* [[CONV21]], align 4 // CHECK10-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64*
// CHECK10-NEXT: [[TMP67:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP67]], align 8
// CHECK10-NEXT: [[TMP68:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK10-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK10-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK10-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i64* // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP69]], align 8
// CHECK10-NEXT: store i64 [[TMP65]], i64* [[TMP70]], align 8 // CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1
// CHECK10-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK10-NEXT: store i8* null, i8** [[TMP70]], align 8
// CHECK10-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i64* // CHECK10-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK10-NEXT: store i64 [[TMP65]], i64* [[TMP72]], align 8 // CHECK10-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32**
// CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP72]], align 8
// CHECK10-NEXT: store i64 4, i64* [[TMP73]], align 8 // CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 0 // CHECK10-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK10-NEXT: store i8* null, i8** [[TMP74]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 8
// CHECK10-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP60]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i64* // CHECK10-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP76]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK10-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 1 // CHECK10-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK10-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i64* // CHECK10-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP78]], align 8 // CHECK10-NEXT: store i64 [[TMP59]], i64* [[TMP77]], align 8
// CHECK10-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 1 // CHECK10-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK10-NEXT: store i64 8, i64* [[TMP79]], align 8 // CHECK10-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP59]], i64* [[TMP79]], align 8
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3
// CHECK10-NEXT: store i8* null, i8** [[TMP80]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP80]], align 8
// CHECK10-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 2 // CHECK10-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32** // CHECK10-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP82]], align 8 // CHECK10-NEXT: [[TMP83:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 2 // CHECK10-NEXT: store i32 [[TMP83]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK10-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 8 // CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP84]], 0
// CHECK10-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 2 // CHECK10-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK10-NEXT: store i64 [[TMP68]], i64* [[TMP85]], align 8 // CHECK10-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK10-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 2 // CHECK10-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP86]], align 8 // CHECK10-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 3 // CHECK10-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP85]], 1
// CHECK10-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK10-NEXT: [[TMP86:%.*]] = zext i32 [[ADD30]] to i64
// CHECK10-NEXT: store i64 [[TMP67]], i64* [[TMP88]], align 8 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP86]])
// CHECK10-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 3 // CHECK10-NEXT: [[TMP87:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP81]], i8** [[TMP82]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK10-NEXT: [[TMP88:%.*]] = icmp ne i32 [[TMP87]], 0
// CHECK10-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK10-NEXT: br i1 [[TMP88]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK10-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 3 // CHECK10: omp_offload.failed31:
// CHECK10-NEXT: store i64 4, i64* [[TMP91]], align 8 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP57]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP59]]) #[[ATTR3]]
// CHECK10-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 3 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK10-NEXT: store i8* null, i8** [[TMP92]], align 8 // CHECK10: omp_offload.cont32:
// CHECK10-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK10-NEXT: [[TMP89:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP89]])
// CHECK10-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0
// CHECK10-NEXT: [[TMP96:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP96]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP97]], 0
// CHECK10-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK10-NEXT: [[SUB31:%.*]] = sub nsw i32 [[DIV30]], 1
// CHECK10-NEXT: store i32 [[SUB31]], i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK10-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK10-NEXT: [[ADD32:%.*]] = add nsw i32 [[TMP98]], 1
// CHECK10-NEXT: [[TMP99:%.*]] = zext i32 [[ADD32]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP99]])
// CHECK10-NEXT: [[TMP100:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP93]], i8** [[TMP94]], i64* [[TMP95]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP101:%.*]] = icmp ne i32 [[TMP100]], 0
// CHECK10-NEXT: br i1 [[TMP101]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK10: omp_offload.failed33:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP65]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP67]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK10: omp_offload.cont34:
// CHECK10-NEXT: [[TMP102:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP102]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP103:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP90:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP103]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP90]])
// CHECK10-NEXT: [[TMP104:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP91:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP104]] // CHECK10-NEXT: ret i32 [[TMP91]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 503 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4 // CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK10-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP20:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP20:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[CONV8:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV8:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP20]], i32* [[CONV8]], align 4, !llvm.access.group !24 // CHECK10-NEXT: store i32 [[TMP20]], i32* [[CONV8]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP21:%.*]] = load i64, i64* [[N_CASTED]], align 8, !llvm.access.group !24 // CHECK10-NEXT: [[TMP21:%.*]] = load i64, i64* [[N_CASTED]], align 8, !llvm.access.group !24
// CHECK10-NEXT: [[TMP22:%.*]] = load i32, i32* [[CONV1]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP22:%.*]] = load i32, i32* [[CONV1]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[CONV9:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV9:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP22]], i32* [[CONV9]], align 4, !llvm.access.group !24 // CHECK10-NEXT: store i32 [[TMP22]], i32* [[CONV9]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !24 // CHECK10-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !24
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP17]], i64 [[TMP19]], i64 [[TMP21]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP23]]), !llvm.access.group !24 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, i32*, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP17]], i64 [[TMP19]], i64 [[TMP21]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP23]]), !llvm.access.group !24
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] // CHECK10-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP25]]
// CHECK10-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK10-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24
Show All 37 Lines
// CHECK10-NEXT: store i32 [[ADD20]], i32* [[I5]], align 4 // CHECK10-NEXT: store i32 [[ADD20]], i32* [[I5]], align 4
// CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK10: .omp.final.done: // CHECK10: .omp.final.done:
// CHECK10-NEXT: br label [[OMP_PRECOND_END]] // CHECK10-NEXT: br label [[OMP_PRECOND_END]]
// CHECK10: omp.precond.end: // CHECK10: omp.precond.end:
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 136 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK10: omp_offload.failed5: // CHECK10: omp_offload.failed5:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK10: omp_offload.cont6: // CHECK10: omp_offload.cont6:
// CHECK10-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK10-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4
Show All 15 Lines
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK10-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK10-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64*
// CHECK10-NEXT: store i64 [[TMP20]], i64* [[TMP29]], align 8 // CHECK10-NEXT: store i64 [[TMP20]], i64* [[TMP29]], align 8
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1 // CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP30]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP30]], align 8
// CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK10-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK10-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK10-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK10: omp_offload.failed11: // CHECK10: omp_offload.failed11:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i64 [[TMP20]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i64 [[TMP20]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK10: omp_offload.cont12: // CHECK10: omp_offload.cont12:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !30 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !30
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !30 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !30
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !30 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !30
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK10: .omp.final.then: // CHECK10: .omp.final.then:
// CHECK10-NEXT: store i32 10, i32* [[I]], align 4 // CHECK10-NEXT: store i32 10, i32* [[I]], align 4
// CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK10: .omp.final.done: // CHECK10: .omp.final.done:
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !36 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !36
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !36 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !36
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !36 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !36
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK10: .omp.final.then: // CHECK10: .omp.final.then:
// CHECK10-NEXT: store i32 10, i32* [[I]], align 4 // CHECK10-NEXT: store i32 10, i32* [[I]], align 4
// CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK10: .omp.final.done: // CHECK10: .omp.final.done:
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 81 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8 // CHECK10-NEXT: store i64 [[DOTCAPTURE_EXPR_]], i64* [[DOTCAPTURE_EXPR__ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__ADDR]] to i32*
// CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP1]], i32* [[CONV1]], align 4 // CHECK10-NEXT: store i32 [[TMP1]], i32* [[CONV1]], align 4
// CHECK10-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP2]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP2]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 39 Lines
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4, !llvm.access.group !42 // CHECK10-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !42 // CHECK10-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !42
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]), !llvm.access.group !42 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]), !llvm.access.group !42
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42
Show All 26 Lines
// CHECK10-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK10-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK10: .omp.final.then: // CHECK10: .omp.final.then:
// CHECK10-NEXT: store i32 10, i32* [[I]], align 4 // CHECK10-NEXT: store i32 10, i32* [[I]], align 4
// CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK10: .omp.final.done: // CHECK10: .omp.final.done:
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 91 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK11-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_16:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4 // CHECK11-NEXT: [[_TMP21:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP23:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: store i32 10, i32* [[M]], align 4 // CHECK11-NEXT: store i32 10, i32* [[M]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK11-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK11-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK11-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK11-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK11-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK11-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK11-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK11-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK11-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK11-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP53]])
// CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK11-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11: omp_offload.failed14:
// CHECK11-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK11-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK11: omp_offload.cont15:
// CHECK11-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: [[TMP56:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: store i32 [[TMP56]], i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK11-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK11-NEXT: store i32 [[TMP57]], i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP61]]) // CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK11-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK11-NEXT: store i32 [[TMP59]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11: omp_offload.failed15: // CHECK11-NEXT: [[TMP61:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK11-NEXT: [[TMP62:%.*]] = sext i32 [[TMP61]] to i64
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK11-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK11: omp_offload.cont16: // CHECK11-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32*
// CHECK11-NEXT: [[TMP64:%.*]] = load i32, i32* [[M]], align 4 // CHECK11-NEXT: store i32 [[TMP58]], i32* [[TMP64]], align 4
// CHECK11-NEXT: store i32 [[TMP64]], i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK11-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK11-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[N_CASTED18]], align 4 // CHECK11-NEXT: store i32 [[TMP58]], i32* [[TMP66]], align 4
// CHECK11-NEXT: [[TMP66:%.*]] = load i32, i32* [[N_CASTED18]], align 4 // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0
// CHECK11-NEXT: [[TMP67:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP67]], align 4
// CHECK11-NEXT: store i32 [[TMP67]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1
// CHECK11-NEXT: [[TMP68:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK11-NEXT: [[TMP69:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP69]], align 4
// CHECK11-NEXT: [[TMP70:%.*]] = sext i32 [[TMP69]] to i64 // CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1
// CHECK11-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK11-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK11-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32* // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP71]], align 4
// CHECK11-NEXT: store i32 [[TMP66]], i32* [[TMP72]], align 4 // CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1
// CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK11-NEXT: store i8* null, i8** [[TMP72]], align 4
// CHECK11-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32* // CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP66]], i32* [[TMP74]], align 4 // CHECK11-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP75]], align 4 // CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2
// CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0 // CHECK11-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK11-NEXT: store i8* null, i8** [[TMP76]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 4
// CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32* // CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP78]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP77]], align 4
// CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1 // CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3
// CHECK11-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK11-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP80]], align 4 // CHECK11-NEXT: store i32 [[TMP60]], i32* [[TMP79]], align 4
// CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 1 // CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3
// CHECK11-NEXT: store i64 4, i64* [[TMP81]], align 4 // CHECK11-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32*
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP60]], i32* [[TMP81]], align 4
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP82]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP82]], align 4
// CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2 // CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK11-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK11-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 4 // CHECK11-NEXT: [[TMP85:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[TMP85]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32** // CHECK11-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP86]], align 4 // CHECK11-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP86]], 0
// CHECK11-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 2 // CHECK11-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK11-NEXT: store i64 [[TMP70]], i64* [[TMP87]], align 4 // CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[DIV25]], 1
// CHECK11-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[SUB26]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP88]], align 4 // CHECK11-NEXT: [[TMP87:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3 // CHECK11-NEXT: [[ADD27:%.*]] = add nsw i32 [[TMP87]], 1
// CHECK11-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK11-NEXT: [[TMP88:%.*]] = zext i32 [[ADD27]] to i64
// CHECK11-NEXT: store i32 [[TMP68]], i32* [[TMP90]], align 4 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP88]])
// CHECK11-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 3 // CHECK11-NEXT: [[TMP89:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP83]], i8** [[TMP84]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32* // CHECK11-NEXT: [[TMP90:%.*]] = icmp ne i32 [[TMP89]], 0
// CHECK11-NEXT: store i32 [[TMP68]], i32* [[TMP92]], align 4 // CHECK11-NEXT: br i1 [[TMP90]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
// CHECK11-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 3 // CHECK11: omp_offload.failed28:
// CHECK11-NEXT: store i64 4, i64* [[TMP93]], align 4 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP58]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP60]]) #[[ATTR3]]
// CHECK11-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT29]]
// CHECK11-NEXT: store i8* null, i8** [[TMP94]], align 4 // CHECK11: omp_offload.cont29:
// CHECK11-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK11-NEXT: [[TMP91:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP91]])
// CHECK11-NEXT: [[TMP97:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
// CHECK11-NEXT: [[TMP98:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP98]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP99]], 0
// CHECK11-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK11-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK11-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK11-NEXT: [[TMP100:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK11-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP100]], 1
// CHECK11-NEXT: [[TMP101:%.*]] = zext i32 [[ADD29]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP101]])
// CHECK11-NEXT: [[TMP102:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP95]], i8** [[TMP96]], i64* [[TMP97]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP103:%.*]] = icmp ne i32 [[TMP102]], 0
// CHECK11-NEXT: br i1 [[TMP103]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK11: omp_offload.failed30:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP66]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP68]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK11: omp_offload.cont31:
// CHECK11-NEXT: [[TMP104:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP104]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP105:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP92:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP105]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP92]])
// CHECK11-NEXT: [[TMP106:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP93:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP106]] // CHECK11-NEXT: ret i32 [[TMP93]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 479 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 64 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[N_ADDR]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[N_ADDR]], align 4, !llvm.access.group !25
// CHECK11-NEXT: store i32 [[TMP18]], i32* [[N_CASTED]], align 4, !llvm.access.group !25 // CHECK11-NEXT: store i32 [[TMP18]], i32* [[N_CASTED]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[N_CASTED]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[N_CASTED]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !25
// CHECK11-NEXT: store i32 [[TMP20]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25 // CHECK11-NEXT: store i32 [[TMP20]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP16]], i32 [[TMP17]], i32 [[TMP19]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP21]]), !llvm.access.group !25 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, i32*, i32)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32 [[TMP16]], i32 [[TMP17]], i32 [[TMP19]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP21]]), !llvm.access.group !25
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP23]] // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
// CHECK11-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25
Show All 37 Lines
// CHECK11-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4 // CHECK11-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4
// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK11: .omp.final.done: // CHECK11: .omp.final.done:
// CHECK11-NEXT: br label [[OMP_PRECOND_END]] // CHECK11-NEXT: br label [[OMP_PRECOND_END]]
// CHECK11: omp.precond.end: // CHECK11: omp.precond.end:
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 131 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK11: omp_offload.failed5: // CHECK11: omp_offload.failed5:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK11: omp_offload.cont6: // CHECK11: omp_offload.cont6:
// CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4
Show All 14 Lines
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK11-NEXT: store i32 [[TMP20]], i32* [[TMP29]], align 4 // CHECK11-NEXT: store i32 [[TMP20]], i32* [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1 // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK11-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK11-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK11-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK11: omp_offload.failed11: // CHECK11: omp_offload.failed11:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i32 [[TMP20]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i32 [[TMP20]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK11: omp_offload.cont12: // CHECK11: omp_offload.cont12:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31
// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !31
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !31 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !31
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !31
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK11-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK11-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK11-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then: // CHECK11: .omp.final.then:
// CHECK11-NEXT: store i32 10, i32* [[I]], align 4 // CHECK11-NEXT: store i32 10, i32* [[I]], align 4
// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK11: .omp.final.done: // CHECK11: .omp.final.done:
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37
// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !37
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !37 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !37
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !37
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP38:![0-9]+]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP38:![0-9]+]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK11-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK11-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK11-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then: // CHECK11: .omp.final.then:
// CHECK11-NEXT: store i32 10, i32* [[I]], align 4 // CHECK11-NEXT: store i32 10, i32* [[I]], align 4
// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK11: .omp.final.done: // CHECK11: .omp.final.done:
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 76 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP2]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP2]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 35 Lines
// CHECK11-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10 // CHECK11-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !43
// CHECK11-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43 // CHECK11-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]), !llvm.access.group !43 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]), !llvm.access.group !43
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43
Show All 26 Lines
// CHECK11-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK11-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then: // CHECK11: .omp.final.then:
// CHECK11-NEXT: store i32 10, i32* [[I]], align 4 // CHECK11-NEXT: store i32 10, i32* [[I]], align 4
// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK11: .omp.final.done: // CHECK11: .omp.final.done:
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 87 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK12-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_16:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4 // CHECK12-NEXT: [[_TMP21:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP23:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: store i32 10, i32* [[M]], align 4 // CHECK12-NEXT: store i32 10, i32* [[M]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK12-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK12-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK12-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK12-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK12-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK12-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK12-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK12-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK12-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK12-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK12-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK12-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK12-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK12-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK12-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK12-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP53]])
// CHECK12-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK12-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12: omp_offload.failed14:
// CHECK12-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK12-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK12: omp_offload.cont15:
// CHECK12-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: [[TMP56:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: store i32 [[TMP56]], i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK12-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK12-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK12-NEXT: store i32 [[TMP57]], i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP61]]) // CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK12-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK12-NEXT: store i32 [[TMP59]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12: omp_offload.failed15: // CHECK12-NEXT: [[TMP61:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK12-NEXT: [[TMP62:%.*]] = sext i32 [[TMP61]] to i64
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK12-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK12: omp_offload.cont16: // CHECK12-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32*
// CHECK12-NEXT: [[TMP64:%.*]] = load i32, i32* [[M]], align 4 // CHECK12-NEXT: store i32 [[TMP58]], i32* [[TMP64]], align 4
// CHECK12-NEXT: store i32 [[TMP64]], i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK12-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK12-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[N_CASTED18]], align 4 // CHECK12-NEXT: store i32 [[TMP58]], i32* [[TMP66]], align 4
// CHECK12-NEXT: [[TMP66:%.*]] = load i32, i32* [[N_CASTED18]], align 4 // CHECK12-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0
// CHECK12-NEXT: [[TMP67:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP67]], align 4
// CHECK12-NEXT: store i32 [[TMP67]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1
// CHECK12-NEXT: [[TMP68:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK12-NEXT: [[TMP69:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP69]], align 4
// CHECK12-NEXT: [[TMP70:%.*]] = sext i32 [[TMP69]] to i64 // CHECK12-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1
// CHECK12-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK12-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK12-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32* // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP71]], align 4
// CHECK12-NEXT: store i32 [[TMP66]], i32* [[TMP72]], align 4 // CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1
// CHECK12-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK12-NEXT: store i8* null, i8** [[TMP72]], align 4
// CHECK12-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32* // CHECK12-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP66]], i32* [[TMP74]], align 4 // CHECK12-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP75]], align 4 // CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2
// CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0 // CHECK12-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK12-NEXT: store i8* null, i8** [[TMP76]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 4
// CHECK12-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32* // CHECK12-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP78]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP77]], align 4
// CHECK12-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1 // CHECK12-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3
// CHECK12-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK12-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP80]], align 4 // CHECK12-NEXT: store i32 [[TMP60]], i32* [[TMP79]], align 4
// CHECK12-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 1 // CHECK12-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3
// CHECK12-NEXT: store i64 4, i64* [[TMP81]], align 4 // CHECK12-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32*
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP60]], i32* [[TMP81]], align 4
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP82]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP82]], align 4
// CHECK12-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2 // CHECK12-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK12-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK12-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 4 // CHECK12-NEXT: [[TMP85:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[TMP85]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32** // CHECK12-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP86]], align 4 // CHECK12-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP86]], 0
// CHECK12-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 2 // CHECK12-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK12-NEXT: store i64 [[TMP70]], i64* [[TMP87]], align 4 // CHECK12-NEXT: [[SUB26:%.*]] = sub nsw i32 [[DIV25]], 1
// CHECK12-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[SUB26]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP88]], align 4 // CHECK12-NEXT: [[TMP87:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3 // CHECK12-NEXT: [[ADD27:%.*]] = add nsw i32 [[TMP87]], 1
// CHECK12-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK12-NEXT: [[TMP88:%.*]] = zext i32 [[ADD27]] to i64
// CHECK12-NEXT: store i32 [[TMP68]], i32* [[TMP90]], align 4 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP88]])
// CHECK12-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 3 // CHECK12-NEXT: [[TMP89:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP83]], i8** [[TMP84]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32* // CHECK12-NEXT: [[TMP90:%.*]] = icmp ne i32 [[TMP89]], 0
// CHECK12-NEXT: store i32 [[TMP68]], i32* [[TMP92]], align 4 // CHECK12-NEXT: br i1 [[TMP90]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
// CHECK12-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 3 // CHECK12: omp_offload.failed28:
// CHECK12-NEXT: store i64 4, i64* [[TMP93]], align 4 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP58]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP60]]) #[[ATTR3]]
// CHECK12-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT29]]
// CHECK12-NEXT: store i8* null, i8** [[TMP94]], align 4 // CHECK12: omp_offload.cont29:
// CHECK12-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK12-NEXT: [[TMP91:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP91]])
// CHECK12-NEXT: [[TMP97:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
// CHECK12-NEXT: [[TMP98:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP98]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP99]], 0
// CHECK12-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK12-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK12-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK12-NEXT: [[TMP100:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK12-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP100]], 1
// CHECK12-NEXT: [[TMP101:%.*]] = zext i32 [[ADD29]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP101]])
// CHECK12-NEXT: [[TMP102:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 4, i8** [[TMP95]], i8** [[TMP96]], i64* [[TMP97]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP103:%.*]] = icmp ne i32 [[TMP102]], 0
// CHECK12-NEXT: br i1 [[TMP103]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK12: omp_offload.failed30:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP66]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP68]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK12: omp_offload.cont31:
// CHECK12-NEXT: [[TMP104:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP104]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP105:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP92:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP105]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP92]])
// CHECK12-NEXT: [[TMP106:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP93:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP106]] // CHECK12-NEXT: ret i32 [[TMP93]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 479 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 64 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[N_ADDR]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[N_ADDR]], align 4, !llvm.access.group !25
// CHECK12-NEXT: store i32 [[TMP18]], i32* [[N_CASTED]], align 4, !llvm.access.group !25 // CHECK12-NEXT: store i32 [[TMP18]], i32* [[N_CASTED]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[N_CASTED]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[N_CASTED]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !25
// CHECK12-NEXT: store i32 [[TMP20]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25 // CHECK12-NEXT: store i32 [[TMP20]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP16]], i32 [[TMP17]], i32 [[TMP19]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP21]]), !llvm.access.group !25 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, i32*, i32)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32 [[TMP16]], i32 [[TMP17]], i32 [[TMP19]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP21]]), !llvm.access.group !25
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP23]] // CHECK12-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
// CHECK12-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 // CHECK12-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25
Show All 37 Lines
// CHECK12-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4 // CHECK12-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4
// CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK12: .omp.final.done: // CHECK12: .omp.final.done:
// CHECK12-NEXT: br label [[OMP_PRECOND_END]] // CHECK12-NEXT: br label [[OMP_PRECOND_END]]
// CHECK12: omp.precond.end: // CHECK12: omp.precond.end:
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 131 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK12: omp_offload.failed5: // CHECK12: omp_offload.failed5:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK12: omp_offload.cont6: // CHECK12: omp_offload.cont6:
// CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: store i32 [[TMP18]], i32* [[DOTCAPTURE_EXPR_]], align 4
Show All 14 Lines
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK12-NEXT: store i32 [[TMP20]], i32* [[TMP29]], align 4 // CHECK12-NEXT: store i32 [[TMP20]], i32* [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1 // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP32:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 2, i8** [[TMP31]], i8** [[TMP32]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK12-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK12-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK12-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK12: omp_offload.failed11: // CHECK12: omp_offload.failed11:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i32 [[TMP20]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]], i32 [[TMP20]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK12: omp_offload.cont12: // CHECK12: omp_offload.cont12:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31
// CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !31
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !31 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !31
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !31
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK12-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK12-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK12-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK12: .omp.final.then: // CHECK12: .omp.final.then:
// CHECK12-NEXT: store i32 10, i32* [[I]], align 4 // CHECK12-NEXT: store i32 10, i32* [[I]], align 4
// CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK12: .omp.final.done: // CHECK12: .omp.final.done:
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37
// CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !37
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !37 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !37
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !37
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP38:![0-9]+]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP38:![0-9]+]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK12-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK12-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK12-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK12: .omp.final.then: // CHECK12: .omp.final.then:
// CHECK12-NEXT: store i32 10, i32* [[I]], align 4 // CHECK12-NEXT: store i32 10, i32* [[I]], align 4
// CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK12: .omp.final.done: // CHECK12: .omp.final.done:
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 76 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: store i32 [[DOTCAPTURE_EXPR_]], i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP2]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP2]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 35 Lines
// CHECK12-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10 // CHECK12-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !43
// CHECK12-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43 // CHECK12-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]), !llvm.access.group !43 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]), !llvm.access.group !43
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43
Show All 26 Lines
// CHECK12-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK12-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK12: .omp.final.then: // CHECK12: .omp.final.then:
// CHECK12-NEXT: store i32 10, i32* [[I]], align 4 // CHECK12-NEXT: store i32 10, i32* [[I]], align 4
// CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK12: .omp.final.done: // CHECK12: .omp.final.done:
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 1,346 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_parallel_for_simd_schedule_codegen.cpp

This file has a very large number of changes (5,472 lines). Show File Contents

clang/test/OpenMP/target_teams_distribute_simd_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 324 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[_TMP19:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP19:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A_CASTED22:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_CASTED22:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS26:%.*]] = alloca [9 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS26:%.*]] = alloca [9 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS27:%.*]] = alloca [9 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS27:%.*]] = alloca [9 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS28:%.*]] = alloca [9 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS28:%.*]] = alloca [9 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 8
// CHECK1-NEXT: [[_TMP29:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP29:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 0, i32* [[A]], align 4 // CHECK1-NEXT: store i32 0, i32* [[A]], align 4
// CHECK1-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK1-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK1-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK1-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
▲ Show 20 Lines • Show All 162 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP93:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK1-NEXT: [[TMP93:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK1-NEXT: [[TMP94:%.*]] = mul nuw i64 [[TMP93]], 8 // CHECK1-NEXT: [[TMP94:%.*]] = mul nuw i64 [[TMP93]], 8
// CHECK1-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0 // CHECK1-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK1-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64* // CHECK1-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64*
// CHECK1-NEXT: store i64 [[TMP90]], i64* [[TMP96]], align 8 // CHECK1-NEXT: store i64 [[TMP90]], i64* [[TMP96]], align 8
// CHECK1-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0 // CHECK1-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK1-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i64* // CHECK1-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i64*
// CHECK1-NEXT: store i64 [[TMP90]], i64* [[TMP98]], align 8 // CHECK1-NEXT: store i64 [[TMP90]], i64* [[TMP98]], align 8
// CHECK1-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK1-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 0
// CHECK1-NEXT: store i64 4, i64* [[TMP99]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP99]], align 8
// CHECK1-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 0 // CHECK1-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 1
// CHECK1-NEXT: store i8* null, i8** [[TMP100]], align 8 // CHECK1-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to [10 x float]**
// CHECK1-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 1 // CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP101]], align 8
// CHECK1-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to [10 x float]** // CHECK1-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 1
// CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP102]], align 8 // CHECK1-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to [10 x float]**
// CHECK1-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 1 // CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP103]], align 8
// CHECK1-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]** // CHECK1-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 1
// CHECK1-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP104]], align 8
// CHECK1-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK1-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 2
// CHECK1-NEXT: store i64 40, i64* [[TMP105]], align 8 // CHECK1-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to i64*
// CHECK1-NEXT: [[TMP106:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 1 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP106]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP106]], align 8 // CHECK1-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 2
// CHECK1-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 2
// CHECK1-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i64* // CHECK1-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i64*
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP108]], align 8 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP108]], align 8
// CHECK1-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 2 // CHECK1-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 2
// CHECK1-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to i64* // CHECK1-NEXT: store i8* null, i8** [[TMP109]], align 8
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP110]], align 8 // CHECK1-NEXT: [[TMP110:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 3
// CHECK1-NEXT: [[TMP111:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK1-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to float**
// CHECK1-NEXT: store i64 8, i64* [[TMP111]], align 8 // CHECK1-NEXT: store float* [[VLA]], float** [[TMP111]], align 8
// CHECK1-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 2 // CHECK1-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 3
// CHECK1-NEXT: store i8* null, i8** [[TMP112]], align 8 // CHECK1-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to float**
// CHECK1-NEXT: [[TMP113:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 3 // CHECK1-NEXT: store float* [[VLA]], float** [[TMP113]], align 8
// CHECK1-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float** // CHECK1-NEXT: store i64 [[TMP92]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 3), align 8
// CHECK1-NEXT: store float* [[VLA]], float** [[TMP114]], align 8 // CHECK1-NEXT: [[TMP114:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 3
// CHECK1-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 3 // CHECK1-NEXT: store i8* null, i8** [[TMP114]], align 8
// CHECK1-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float** // CHECK1-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 4
// CHECK1-NEXT: store float* [[VLA]], float** [[TMP116]], align 8 // CHECK1-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]**
// CHECK1-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 8
// CHECK1-NEXT: store i64 [[TMP92]], i64* [[TMP117]], align 8 // CHECK1-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 4
// CHECK1-NEXT: [[TMP118:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 3 // CHECK1-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to [5 x [10 x double]]**
// CHECK1-NEXT: store i8* null, i8** [[TMP118]], align 8 // CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP118]], align 8
// CHECK1-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 4 // CHECK1-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 4
// CHECK1-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to [5 x [10 x double]]** // CHECK1-NEXT: store i8* null, i8** [[TMP119]], align 8
// CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP120]], align 8 // CHECK1-NEXT: [[TMP120:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 5
// CHECK1-NEXT: [[TMP121:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 4 // CHECK1-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i64*
// CHECK1-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to [5 x [10 x double]]** // CHECK1-NEXT: store i64 5, i64* [[TMP121]], align 8
// CHECK1-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP122]], align 8 // CHECK1-NEXT: [[TMP122:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 5
// CHECK1-NEXT: [[TMP123:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK1-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i64*
// CHECK1-NEXT: store i64 400, i64* [[TMP123]], align 8 // CHECK1-NEXT: store i64 5, i64* [[TMP123]], align 8
// CHECK1-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 4 // CHECK1-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 5
// CHECK1-NEXT: store i8* null, i8** [[TMP124]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP124]], align 8
// CHECK1-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 5 // CHECK1-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 6
// CHECK1-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64* // CHECK1-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64*
// CHECK1-NEXT: store i64 5, i64* [[TMP126]], align 8 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP126]], align 8
// CHECK1-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 5 // CHECK1-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 6
// CHECK1-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64* // CHECK1-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64*
// CHECK1-NEXT: store i64 5, i64* [[TMP128]], align 8 // CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP128]], align 8
// CHECK1-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK1-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 6
// CHECK1-NEXT: store i64 8, i64* [[TMP129]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP129]], align 8
// CHECK1-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 5 // CHECK1-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 7
// CHECK1-NEXT: store i8* null, i8** [[TMP130]], align 8 // CHECK1-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to double**
// CHECK1-NEXT: [[TMP131:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 6 // CHECK1-NEXT: store double* [[VLA1]], double** [[TMP131]], align 8
// CHECK1-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to i64* // CHECK1-NEXT: [[TMP132:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 7
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP132]], align 8 // CHECK1-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to double**
// CHECK1-NEXT: [[TMP133:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 6 // CHECK1-NEXT: store double* [[VLA1]], double** [[TMP133]], align 8
// CHECK1-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to i64* // CHECK1-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 7), align 8
// CHECK1-NEXT: store i64 [[TMP5]], i64* [[TMP134]], align 8 // CHECK1-NEXT: [[TMP134:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 7
// CHECK1-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK1-NEXT: store i8* null, i8** [[TMP134]], align 8
// CHECK1-NEXT: store i64 8, i64* [[TMP135]], align 8 // CHECK1-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 8
// CHECK1-NEXT: [[TMP136:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 6 // CHECK1-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT**
// CHECK1-NEXT: store i8* null, i8** [[TMP136]], align 8 // CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 8
// CHECK1-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 7 // CHECK1-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 8
// CHECK1-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to double** // CHECK1-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT**
// CHECK1-NEXT: store double* [[VLA1]], double** [[TMP138]], align 8 // CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 8
// CHECK1-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 7 // CHECK1-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 8
// CHECK1-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to double** // CHECK1-NEXT: store i8* null, i8** [[TMP139]], align 8
// CHECK1-NEXT: store double* [[VLA1]], double** [[TMP140]], align 8 // CHECK1-NEXT: [[TMP140:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK1-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK1-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP94]], i64* [[TMP141]], align 8
// CHECK1-NEXT: [[TMP142:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 7
// CHECK1-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK1-NEXT: [[TMP143:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 8
// CHECK1-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to %struct.TT**
// CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP144]], align 8
// CHECK1-NEXT: [[TMP145:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 8
// CHECK1-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to %struct.TT**
// CHECK1-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP146]], align 8
// CHECK1-NEXT: [[TMP147:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK1-NEXT: store i64 16, i64* [[TMP147]], align 8
// CHECK1-NEXT: [[TMP148:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 8
// CHECK1-NEXT: store i8* null, i8** [[TMP148]], align 8
// CHECK1-NEXT: [[TMP149:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK1-NEXT: [[TMP150:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK1-NEXT: [[TMP151:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK1-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK1-NEXT: [[TMP142:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP140]], i8** [[TMP141]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK1-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0 // CHECK1-NEXT: [[TMP143:%.*]] = icmp ne i32 [[TMP142]], 0
// CHECK1-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]] // CHECK1-NEXT: br i1 [[TMP143]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK1: omp_offload.failed30: // CHECK1: omp_offload.failed30:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT31]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK1: omp_offload.cont31: // CHECK1: omp_offload.cont31:
// CHECK1-NEXT: br label [[OMP_IF_END33:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END33:%.*]]
// CHECK1: omp_if.else32: // CHECK1: omp_if.else32:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_IF_END33]] // CHECK1-NEXT: br label [[OMP_IF_END33]]
// CHECK1: omp_if.end33: // CHECK1: omp_if.end33:
// CHECK1-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK1-NEXT: [[TMP144:%.*]] = load i32, i32* [[A]], align 4
// CHECK1-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP145:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP145]])
// CHECK1-NEXT: ret i32 [[TMP154]] // CHECK1-NEXT: ret i32 [[TMP144]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97
// CHECK1-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK1-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 685 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK1-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK1-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK1-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 15 Lines
// CHECK1-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK1-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK1-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK1-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK1-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK1-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK1-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK1-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK1-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK1-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK1-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK1-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK1-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK1-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK1-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK1-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK1-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK1-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK1-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK1-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK1-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 8
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK1-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK1-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK1-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK1-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK1-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK1-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK1-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK1-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_IF_END]] // CHECK1-NEXT: br label [[OMP_IF_END]]
// CHECK1: omp_if.end: // CHECK1: omp_if.end:
// CHECK1-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK1-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK1-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK1-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK1-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK1-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK1-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK1-NEXT: ret i32 [[ADD4]] // CHECK1-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK1-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK1-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1
// CHECK1-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK1-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK1-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1 // CHECK1-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1
// CHECK1-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK1-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK1-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK1-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK1-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1 // CHECK1-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1
// CHECK1-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64 // CHECK1-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK1-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK1-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK1-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK1-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK1-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK1-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK1-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed: // CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: br label [[OMP_IF_END:%.*]] // CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK1: omp_if.else: // CHECK1: omp_if.else:
Show All 22 Lines
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK1-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 117 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2 // CHECK1-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2
// CHECK1-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK1-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2 // CHECK1-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2
// CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK1-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK1-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK1-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK1-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1 // CHECK1-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1
// CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[N:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[N:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 145 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK1-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK1-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK1-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: [[_TMP19:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[_TMP19:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A_CASTED22:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_CASTED22:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS26:%.*]] = alloca [9 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS26:%.*]] = alloca [9 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS27:%.*]] = alloca [9 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS27:%.*]] = alloca [9 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS28:%.*]] = alloca [9 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS28:%.*]] = alloca [9 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 8
// CHECK2-NEXT: [[_TMP29:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[_TMP29:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 0, i32* [[A]], align 4 // CHECK2-NEXT: store i32 0, i32* [[A]], align 4
// CHECK2-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK2-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK2-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK2-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK2-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
▲ Show 20 Lines • Show All 162 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP93:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK2-NEXT: [[TMP93:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK2-NEXT: [[TMP94:%.*]] = mul nuw i64 [[TMP93]], 8 // CHECK2-NEXT: [[TMP94:%.*]] = mul nuw i64 [[TMP93]], 8
// CHECK2-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0 // CHECK2-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK2-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64* // CHECK2-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64*
// CHECK2-NEXT: store i64 [[TMP90]], i64* [[TMP96]], align 8 // CHECK2-NEXT: store i64 [[TMP90]], i64* [[TMP96]], align 8
// CHECK2-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0 // CHECK2-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK2-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i64* // CHECK2-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i64*
// CHECK2-NEXT: store i64 [[TMP90]], i64* [[TMP98]], align 8 // CHECK2-NEXT: store i64 [[TMP90]], i64* [[TMP98]], align 8
// CHECK2-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK2-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 0
// CHECK2-NEXT: store i64 4, i64* [[TMP99]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP99]], align 8
// CHECK2-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 0 // CHECK2-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP100]], align 8 // CHECK2-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to [10 x float]**
// CHECK2-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 1 // CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP101]], align 8
// CHECK2-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to [10 x float]** // CHECK2-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 1
// CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP102]], align 8 // CHECK2-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to [10 x float]**
// CHECK2-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 1 // CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP103]], align 8
// CHECK2-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]** // CHECK2-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 1
// CHECK2-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP104]], align 8
// CHECK2-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK2-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 2
// CHECK2-NEXT: store i64 40, i64* [[TMP105]], align 8 // CHECK2-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to i64*
// CHECK2-NEXT: [[TMP106:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 1 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP106]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP106]], align 8 // CHECK2-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 2
// CHECK2-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 2
// CHECK2-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i64* // CHECK2-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i64*
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP108]], align 8 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP108]], align 8
// CHECK2-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 2 // CHECK2-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 2
// CHECK2-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to i64* // CHECK2-NEXT: store i8* null, i8** [[TMP109]], align 8
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP110]], align 8 // CHECK2-NEXT: [[TMP110:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 3
// CHECK2-NEXT: [[TMP111:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK2-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to float**
// CHECK2-NEXT: store i64 8, i64* [[TMP111]], align 8 // CHECK2-NEXT: store float* [[VLA]], float** [[TMP111]], align 8
// CHECK2-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 2 // CHECK2-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 3
// CHECK2-NEXT: store i8* null, i8** [[TMP112]], align 8 // CHECK2-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to float**
// CHECK2-NEXT: [[TMP113:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 3 // CHECK2-NEXT: store float* [[VLA]], float** [[TMP113]], align 8
// CHECK2-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float** // CHECK2-NEXT: store i64 [[TMP92]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 3), align 8
// CHECK2-NEXT: store float* [[VLA]], float** [[TMP114]], align 8 // CHECK2-NEXT: [[TMP114:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 3
// CHECK2-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 3 // CHECK2-NEXT: store i8* null, i8** [[TMP114]], align 8
// CHECK2-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float** // CHECK2-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 4
// CHECK2-NEXT: store float* [[VLA]], float** [[TMP116]], align 8 // CHECK2-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]**
// CHECK2-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 8
// CHECK2-NEXT: store i64 [[TMP92]], i64* [[TMP117]], align 8 // CHECK2-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 4
// CHECK2-NEXT: [[TMP118:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 3 // CHECK2-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to [5 x [10 x double]]**
// CHECK2-NEXT: store i8* null, i8** [[TMP118]], align 8 // CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP118]], align 8
// CHECK2-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 4 // CHECK2-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 4
// CHECK2-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to [5 x [10 x double]]** // CHECK2-NEXT: store i8* null, i8** [[TMP119]], align 8
// CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP120]], align 8 // CHECK2-NEXT: [[TMP120:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 5
// CHECK2-NEXT: [[TMP121:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 4 // CHECK2-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i64*
// CHECK2-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to [5 x [10 x double]]** // CHECK2-NEXT: store i64 5, i64* [[TMP121]], align 8
// CHECK2-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP122]], align 8 // CHECK2-NEXT: [[TMP122:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 5
// CHECK2-NEXT: [[TMP123:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK2-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i64*
// CHECK2-NEXT: store i64 400, i64* [[TMP123]], align 8 // CHECK2-NEXT: store i64 5, i64* [[TMP123]], align 8
// CHECK2-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 4 // CHECK2-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 5
// CHECK2-NEXT: store i8* null, i8** [[TMP124]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP124]], align 8
// CHECK2-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 5 // CHECK2-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 6
// CHECK2-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64* // CHECK2-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64*
// CHECK2-NEXT: store i64 5, i64* [[TMP126]], align 8 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP126]], align 8
// CHECK2-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 5 // CHECK2-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 6
// CHECK2-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64* // CHECK2-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64*
// CHECK2-NEXT: store i64 5, i64* [[TMP128]], align 8 // CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP128]], align 8
// CHECK2-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK2-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 6
// CHECK2-NEXT: store i64 8, i64* [[TMP129]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP129]], align 8
// CHECK2-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 5 // CHECK2-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 7
// CHECK2-NEXT: store i8* null, i8** [[TMP130]], align 8 // CHECK2-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to double**
// CHECK2-NEXT: [[TMP131:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 6 // CHECK2-NEXT: store double* [[VLA1]], double** [[TMP131]], align 8
// CHECK2-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to i64* // CHECK2-NEXT: [[TMP132:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 7
// CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP132]], align 8 // CHECK2-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to double**
// CHECK2-NEXT: [[TMP133:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 6 // CHECK2-NEXT: store double* [[VLA1]], double** [[TMP133]], align 8
// CHECK2-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to i64* // CHECK2-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 7), align 8
// CHECK2-NEXT: store i64 [[TMP5]], i64* [[TMP134]], align 8 // CHECK2-NEXT: [[TMP134:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 7
// CHECK2-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK2-NEXT: store i8* null, i8** [[TMP134]], align 8
// CHECK2-NEXT: store i64 8, i64* [[TMP135]], align 8 // CHECK2-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 8
// CHECK2-NEXT: [[TMP136:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 6 // CHECK2-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT**
// CHECK2-NEXT: store i8* null, i8** [[TMP136]], align 8 // CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 8
// CHECK2-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 7 // CHECK2-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 8
// CHECK2-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to double** // CHECK2-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT**
// CHECK2-NEXT: store double* [[VLA1]], double** [[TMP138]], align 8 // CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 8
// CHECK2-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 7 // CHECK2-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 8
// CHECK2-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to double** // CHECK2-NEXT: store i8* null, i8** [[TMP139]], align 8
// CHECK2-NEXT: store double* [[VLA1]], double** [[TMP140]], align 8 // CHECK2-NEXT: [[TMP140:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK2-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK2-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK2-NEXT: store i64 [[TMP94]], i64* [[TMP141]], align 8
// CHECK2-NEXT: [[TMP142:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 7
// CHECK2-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK2-NEXT: [[TMP143:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 8
// CHECK2-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to %struct.TT**
// CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP144]], align 8
// CHECK2-NEXT: [[TMP145:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 8
// CHECK2-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to %struct.TT**
// CHECK2-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP146]], align 8
// CHECK2-NEXT: [[TMP147:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK2-NEXT: store i64 16, i64* [[TMP147]], align 8
// CHECK2-NEXT: [[TMP148:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 8
// CHECK2-NEXT: store i8* null, i8** [[TMP148]], align 8
// CHECK2-NEXT: [[TMP149:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK2-NEXT: [[TMP150:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK2-NEXT: [[TMP151:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK2-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK2-NEXT: [[TMP142:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP140]], i8** [[TMP141]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK2-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0 // CHECK2-NEXT: [[TMP143:%.*]] = icmp ne i32 [[TMP142]], 0
// CHECK2-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]] // CHECK2-NEXT: br i1 [[TMP143]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK2: omp_offload.failed30: // CHECK2: omp_offload.failed30:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT31]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK2: omp_offload.cont31: // CHECK2: omp_offload.cont31:
// CHECK2-NEXT: br label [[OMP_IF_END33:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END33:%.*]]
// CHECK2: omp_if.else32: // CHECK2: omp_if.else32:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_IF_END33]] // CHECK2-NEXT: br label [[OMP_IF_END33]]
// CHECK2: omp_if.end33: // CHECK2: omp_if.end33:
// CHECK2-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK2-NEXT: [[TMP144:%.*]] = load i32, i32* [[A]], align 4
// CHECK2-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: [[TMP145:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP145]])
// CHECK2-NEXT: ret i32 [[TMP154]] // CHECK2-NEXT: ret i32 [[TMP144]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97
// CHECK2-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK2-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 685 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK2-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK2-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK2-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 15 Lines
// CHECK2-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK2-NEXT: [[TMP8:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2 // CHECK2-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 2
// CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK2-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8 // CHECK2-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK2-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK2-NEXT: store double* [[A]], double** [[TMP13]], align 8 // CHECK2-NEXT: store double* [[A]], double** [[TMP13]], align 8
// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK2-NEXT: store i64 8, i64* [[TMP14]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK2-NEXT: store i8* null, i8** [[TMP15]], align 8 // CHECK2-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP19]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: store i64 4, i64* [[TMP20]], align 8 // CHECK2-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK2-NEXT: store i64 2, i64* [[TMP21]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP21]], align 8 // CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK2-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK2-NEXT: store i64 2, i64* [[TMP23]], align 8 // CHECK2-NEXT: store i64 2, i64* [[TMP23]], align 8
// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK2-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK2-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK2-NEXT: store i64 2, i64* [[TMP25]], align 8 // CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK2-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK2-NEXT: store i64 8, i64* [[TMP26]], align 8 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP26]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK2-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK2-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP28]], align 8
// CHECK2-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK2-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK2-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK2-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8 // CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 8
// CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK2-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK2-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK2-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 8
// CHECK2-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK2-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 8
// CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK2-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK2-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 8 // CHECK2-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK2-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 8
// CHECK2-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK2-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 8
// CHECK2-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK2-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK2-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK2-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK2-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK2-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK2-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK2-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP6]], i64 2, i64 [[TMP2]], i16* [[VLA]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_IF_END]] // CHECK2-NEXT: br label [[OMP_IF_END]]
// CHECK2: omp_if.end: // CHECK2: omp_if.end:
// CHECK2-NEXT: [[TMP45:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK2-NEXT: [[TMP39:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP45]] // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP39]]
// CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK2-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK2-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP46]] to i32 // CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP40]] to i32
// CHECK2-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK2-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP47]] // CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], [[TMP41]]
// CHECK2-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK2-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK2-NEXT: ret i32 [[ADD4]] // CHECK2-NEXT: ret i32 [[ADD4]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK2-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK2-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1
// CHECK2-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK2-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK2-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1 // CHECK2-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1
// CHECK2-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK2-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK2-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK2-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK2-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1 // CHECK2-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1
// CHECK2-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64 // CHECK2-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK2-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK2-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK2-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK2-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK2-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK2-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK2-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK2-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK2-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK2-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK2-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK2-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK2-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK2-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed: // CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont: // CHECK2: omp_offload.cont:
// CHECK2-NEXT: br label [[OMP_IF_END:%.*]] // CHECK2-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK2: omp_if.else: // CHECK2: omp_if.else:
Show All 22 Lines
// CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32* // CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32* // CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK2-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK2-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 117 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2 // CHECK2-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2
// CHECK2-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK2-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK2-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2 // CHECK2-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2
// CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK2-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK2-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK2-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK2-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK2-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1 // CHECK2-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1
// CHECK2-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK2-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[N:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[N:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 145 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK2-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK2-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK2-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK2-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK2-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS12:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS12:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS13:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS13:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4
// CHECK3-NEXT: [[_TMP15:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[_TMP15:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A_CASTED18:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_CASTED18:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [9 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [9 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [9 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [9 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [9 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [9 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 4
// CHECK3-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[A]], align 4 // CHECK3-NEXT: store i32 0, i32* [[A]], align 4
// CHECK3-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK3-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK3-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK3-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
▲ Show 20 Lines • Show All 157 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP93:%.*]] = mul nuw i32 [[TMP92]], 8 // CHECK3-NEXT: [[TMP93:%.*]] = mul nuw i32 [[TMP92]], 8
// CHECK3-NEXT: [[TMP94:%.*]] = sext i32 [[TMP93]] to i64 // CHECK3-NEXT: [[TMP94:%.*]] = sext i32 [[TMP93]] to i64
// CHECK3-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 // CHECK3-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK3-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i32* // CHECK3-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i32*
// CHECK3-NEXT: store i32 [[TMP88]], i32* [[TMP96]], align 4 // CHECK3-NEXT: store i32 [[TMP88]], i32* [[TMP96]], align 4
// CHECK3-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 // CHECK3-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK3-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32* // CHECK3-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32*
// CHECK3-NEXT: store i32 [[TMP88]], i32* [[TMP98]], align 4 // CHECK3-NEXT: store i32 [[TMP88]], i32* [[TMP98]], align 4
// CHECK3-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK3-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 0
// CHECK3-NEXT: store i64 4, i64* [[TMP99]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP99]], align 4
// CHECK3-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 0 // CHECK3-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP100]], align 4 // CHECK3-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to [10 x float]**
// CHECK3-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1 // CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP101]], align 4
// CHECK3-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to [10 x float]** // CHECK3-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP102]], align 4 // CHECK3-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to [10 x float]**
// CHECK3-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1 // CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP103]], align 4
// CHECK3-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]** // CHECK3-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 1
// CHECK3-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP104]], align 4
// CHECK3-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK3-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK3-NEXT: store i64 40, i64* [[TMP105]], align 4 // CHECK3-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to i32*
// CHECK3-NEXT: [[TMP106:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP106]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP106]], align 4 // CHECK3-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK3-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK3-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i32* // CHECK3-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP108]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP108]], align 4
// CHECK3-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2 // CHECK3-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 2
// CHECK3-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to i32* // CHECK3-NEXT: store i8* null, i8** [[TMP109]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP110]], align 4 // CHECK3-NEXT: [[TMP110:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK3-NEXT: [[TMP111:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK3-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to float**
// CHECK3-NEXT: store i64 4, i64* [[TMP111]], align 4 // CHECK3-NEXT: store float* [[VLA]], float** [[TMP111]], align 4
// CHECK3-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 2 // CHECK3-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP112]], align 4 // CHECK3-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to float**
// CHECK3-NEXT: [[TMP113:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3 // CHECK3-NEXT: store float* [[VLA]], float** [[TMP113]], align 4
// CHECK3-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float** // CHECK3-NEXT: store i64 [[TMP91]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 3), align 4
// CHECK3-NEXT: store float* [[VLA]], float** [[TMP114]], align 4 // CHECK3-NEXT: [[TMP114:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 3
// CHECK3-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3 // CHECK3-NEXT: store i8* null, i8** [[TMP114]], align 4
// CHECK3-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float** // CHECK3-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 4
// CHECK3-NEXT: store float* [[VLA]], float** [[TMP116]], align 4 // CHECK3-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]**
// CHECK3-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 4
// CHECK3-NEXT: store i64 [[TMP91]], i64* [[TMP117]], align 4 // CHECK3-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 4
// CHECK3-NEXT: [[TMP118:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 3 // CHECK3-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to [5 x [10 x double]]**
// CHECK3-NEXT: store i8* null, i8** [[TMP118]], align 4 // CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP118]], align 4
// CHECK3-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 4 // CHECK3-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 4
// CHECK3-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to [5 x [10 x double]]** // CHECK3-NEXT: store i8* null, i8** [[TMP119]], align 4
// CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP120]], align 4 // CHECK3-NEXT: [[TMP120:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 5
// CHECK3-NEXT: [[TMP121:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 4 // CHECK3-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i32*
// CHECK3-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to [5 x [10 x double]]** // CHECK3-NEXT: store i32 5, i32* [[TMP121]], align 4
// CHECK3-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP122]], align 4 // CHECK3-NEXT: [[TMP122:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 5
// CHECK3-NEXT: [[TMP123:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK3-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i32*
// CHECK3-NEXT: store i64 400, i64* [[TMP123]], align 4 // CHECK3-NEXT: store i32 5, i32* [[TMP123]], align 4
// CHECK3-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 4 // CHECK3-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 5
// CHECK3-NEXT: store i8* null, i8** [[TMP124]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP124]], align 4
// CHECK3-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 5 // CHECK3-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 6
// CHECK3-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32* // CHECK3-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32*
// CHECK3-NEXT: store i32 5, i32* [[TMP126]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP126]], align 4
// CHECK3-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 5 // CHECK3-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 6
// CHECK3-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i32* // CHECK3-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i32*
// CHECK3-NEXT: store i32 5, i32* [[TMP128]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP128]], align 4
// CHECK3-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK3-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 6
// CHECK3-NEXT: store i64 4, i64* [[TMP129]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP129]], align 4
// CHECK3-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 5 // CHECK3-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 7
// CHECK3-NEXT: store i8* null, i8** [[TMP130]], align 4 // CHECK3-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to double**
// CHECK3-NEXT: [[TMP131:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 6 // CHECK3-NEXT: store double* [[VLA1]], double** [[TMP131]], align 4
// CHECK3-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to i32* // CHECK3-NEXT: [[TMP132:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 7
// CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP132]], align 4 // CHECK3-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to double**
// CHECK3-NEXT: [[TMP133:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 6 // CHECK3-NEXT: store double* [[VLA1]], double** [[TMP133]], align 4
// CHECK3-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to i32* // CHECK3-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 7), align 4
// CHECK3-NEXT: store i32 [[TMP3]], i32* [[TMP134]], align 4 // CHECK3-NEXT: [[TMP134:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 7
// CHECK3-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK3-NEXT: store i8* null, i8** [[TMP134]], align 4
// CHECK3-NEXT: store i64 4, i64* [[TMP135]], align 4 // CHECK3-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 8
// CHECK3-NEXT: [[TMP136:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 6 // CHECK3-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT**
// CHECK3-NEXT: store i8* null, i8** [[TMP136]], align 4 // CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 4
// CHECK3-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 7 // CHECK3-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 8
// CHECK3-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to double** // CHECK3-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT**
// CHECK3-NEXT: store double* [[VLA1]], double** [[TMP138]], align 4 // CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 4
// CHECK3-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 7 // CHECK3-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 8
// CHECK3-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to double** // CHECK3-NEXT: store i8* null, i8** [[TMP139]], align 4
// CHECK3-NEXT: store double* [[VLA1]], double** [[TMP140]], align 4 // CHECK3-NEXT: [[TMP140:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK3-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK3-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK3-NEXT: store i64 [[TMP94]], i64* [[TMP141]], align 4
// CHECK3-NEXT: [[TMP142:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 7
// CHECK3-NEXT: store i8* null, i8** [[TMP142]], align 4
// CHECK3-NEXT: [[TMP143:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 8
// CHECK3-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to %struct.TT**
// CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP144]], align 4
// CHECK3-NEXT: [[TMP145:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 8
// CHECK3-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to %struct.TT**
// CHECK3-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP146]], align 4
// CHECK3-NEXT: [[TMP147:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK3-NEXT: store i64 12, i64* [[TMP147]], align 4
// CHECK3-NEXT: [[TMP148:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 8
// CHECK3-NEXT: store i8* null, i8** [[TMP148]], align 4
// CHECK3-NEXT: [[TMP149:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK3-NEXT: [[TMP150:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK3-NEXT: [[TMP151:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK3-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK3-NEXT: [[TMP142:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP140]], i8** [[TMP141]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK3-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0 // CHECK3-NEXT: [[TMP143:%.*]] = icmp ne i32 [[TMP142]], 0
// CHECK3-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED25:%.*]], label [[OMP_OFFLOAD_CONT26:%.*]] // CHECK3-NEXT: br i1 [[TMP143]], label [[OMP_OFFLOAD_FAILED25:%.*]], label [[OMP_OFFLOAD_CONT26:%.*]]
// CHECK3: omp_offload.failed25: // CHECK3: omp_offload.failed25:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT26]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT26]]
// CHECK3: omp_offload.cont26: // CHECK3: omp_offload.cont26:
// CHECK3-NEXT: br label [[OMP_IF_END28:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END28:%.*]]
// CHECK3: omp_if.else27: // CHECK3: omp_if.else27:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_IF_END28]] // CHECK3-NEXT: br label [[OMP_IF_END28]]
// CHECK3: omp_if.end28: // CHECK3: omp_if.end28:
// CHECK3-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK3-NEXT: [[TMP144:%.*]] = load i32, i32* [[A]], align 4
// CHECK3-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: [[TMP145:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP145]])
// CHECK3-NEXT: ret i32 [[TMP154]] // CHECK3-NEXT: ret i32 [[TMP144]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97
// CHECK3-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK3-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 672 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK3-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK3-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK3-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 14 Lines
// CHECK3-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK3-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK3-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK3-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK3-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK3-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK3-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK3-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK3-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK3-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK3-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK3-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK3-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK3-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK3-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK3-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK3-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK3-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK3-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK3-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK3-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK3-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK3-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 4
// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK3-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK3-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4
// CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK3-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK3-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK3-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK3-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK3-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK3-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK3-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_IF_END]] // CHECK3-NEXT: br label [[OMP_IF_END]]
// CHECK3: omp_if.end: // CHECK3: omp_if.end:
// CHECK3-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK3-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK3-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK3-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK3-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK3-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK3-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK3-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK3-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK3-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK3-NEXT: ret i32 [[ADD3]] // CHECK3-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK3-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK3-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1 // CHECK3-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1
// CHECK3-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK3-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK3-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1 // CHECK3-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1
// CHECK3-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK3-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK3-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK3-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK3-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1 // CHECK3-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1
// CHECK3-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64 // CHECK3-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64
// CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK3-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK3-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK3-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK3-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK3-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
▲ Show 20 Lines • Show All 53 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK3-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK3-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK3-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed: // CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: br label [[OMP_IF_END:%.*]] // CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK3: omp_if.else: // CHECK3: omp_if.else:
Show All 20 Lines
// CHECK3-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK3-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 112 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK3-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK3-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK3-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2 // CHECK3-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK3-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK3-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK3-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK3-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1 // CHECK3-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[N:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[N:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 141 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK3-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK3-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK3-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK3-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 114 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS12:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS12:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS13:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS13:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4
// CHECK4-NEXT: [[_TMP15:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[_TMP15:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[A_CASTED18:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_CASTED18:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [9 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [9 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [9 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [9 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [9 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [9 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 4
// CHECK4-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK4-NEXT: store i32 0, i32* [[A]], align 4 // CHECK4-NEXT: store i32 0, i32* [[A]], align 4
// CHECK4-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK4-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK4-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK4-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
▲ Show 20 Lines • Show All 157 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP93:%.*]] = mul nuw i32 [[TMP92]], 8 // CHECK4-NEXT: [[TMP93:%.*]] = mul nuw i32 [[TMP92]], 8
// CHECK4-NEXT: [[TMP94:%.*]] = sext i32 [[TMP93]] to i64 // CHECK4-NEXT: [[TMP94:%.*]] = sext i32 [[TMP93]] to i64
// CHECK4-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 // CHECK4-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK4-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i32* // CHECK4-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i32*
// CHECK4-NEXT: store i32 [[TMP88]], i32* [[TMP96]], align 4 // CHECK4-NEXT: store i32 [[TMP88]], i32* [[TMP96]], align 4
// CHECK4-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 // CHECK4-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK4-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32* // CHECK4-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32*
// CHECK4-NEXT: store i32 [[TMP88]], i32* [[TMP98]], align 4 // CHECK4-NEXT: store i32 [[TMP88]], i32* [[TMP98]], align 4
// CHECK4-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK4-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 0
// CHECK4-NEXT: store i64 4, i64* [[TMP99]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP99]], align 4
// CHECK4-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 0 // CHECK4-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK4-NEXT: store i8* null, i8** [[TMP100]], align 4 // CHECK4-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to [10 x float]**
// CHECK4-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1 // CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP101]], align 4
// CHECK4-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to [10 x float]** // CHECK4-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP102]], align 4 // CHECK4-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to [10 x float]**
// CHECK4-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1 // CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP103]], align 4
// CHECK4-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]** // CHECK4-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 1
// CHECK4-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP104]], align 4
// CHECK4-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK4-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK4-NEXT: store i64 40, i64* [[TMP105]], align 4 // CHECK4-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to i32*
// CHECK4-NEXT: [[TMP106:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP106]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP106]], align 4 // CHECK4-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK4-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK4-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i32* // CHECK4-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i32*
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP108]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP108]], align 4
// CHECK4-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2 // CHECK4-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 2
// CHECK4-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to i32* // CHECK4-NEXT: store i8* null, i8** [[TMP109]], align 4
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP110]], align 4 // CHECK4-NEXT: [[TMP110:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK4-NEXT: [[TMP111:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK4-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to float**
// CHECK4-NEXT: store i64 4, i64* [[TMP111]], align 4 // CHECK4-NEXT: store float* [[VLA]], float** [[TMP111]], align 4
// CHECK4-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 2 // CHECK4-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP112]], align 4 // CHECK4-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to float**
// CHECK4-NEXT: [[TMP113:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3 // CHECK4-NEXT: store float* [[VLA]], float** [[TMP113]], align 4
// CHECK4-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float** // CHECK4-NEXT: store i64 [[TMP91]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 3), align 4
// CHECK4-NEXT: store float* [[VLA]], float** [[TMP114]], align 4 // CHECK4-NEXT: [[TMP114:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 3
// CHECK4-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3 // CHECK4-NEXT: store i8* null, i8** [[TMP114]], align 4
// CHECK4-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float** // CHECK4-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 4
// CHECK4-NEXT: store float* [[VLA]], float** [[TMP116]], align 4 // CHECK4-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]**
// CHECK4-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 4
// CHECK4-NEXT: store i64 [[TMP91]], i64* [[TMP117]], align 4 // CHECK4-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 4
// CHECK4-NEXT: [[TMP118:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 3 // CHECK4-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to [5 x [10 x double]]**
// CHECK4-NEXT: store i8* null, i8** [[TMP118]], align 4 // CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP118]], align 4
// CHECK4-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 4 // CHECK4-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 4
// CHECK4-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to [5 x [10 x double]]** // CHECK4-NEXT: store i8* null, i8** [[TMP119]], align 4
// CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP120]], align 4 // CHECK4-NEXT: [[TMP120:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 5
// CHECK4-NEXT: [[TMP121:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 4 // CHECK4-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i32*
// CHECK4-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to [5 x [10 x double]]** // CHECK4-NEXT: store i32 5, i32* [[TMP121]], align 4
// CHECK4-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP122]], align 4 // CHECK4-NEXT: [[TMP122:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 5
// CHECK4-NEXT: [[TMP123:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK4-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i32*
// CHECK4-NEXT: store i64 400, i64* [[TMP123]], align 4 // CHECK4-NEXT: store i32 5, i32* [[TMP123]], align 4
// CHECK4-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 4 // CHECK4-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 5
// CHECK4-NEXT: store i8* null, i8** [[TMP124]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP124]], align 4
// CHECK4-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 5 // CHECK4-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 6
// CHECK4-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32* // CHECK4-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32*
// CHECK4-NEXT: store i32 5, i32* [[TMP126]], align 4 // CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP126]], align 4
// CHECK4-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 5 // CHECK4-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 6
// CHECK4-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i32* // CHECK4-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i32*
// CHECK4-NEXT: store i32 5, i32* [[TMP128]], align 4 // CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP128]], align 4
// CHECK4-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK4-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 6
// CHECK4-NEXT: store i64 4, i64* [[TMP129]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP129]], align 4
// CHECK4-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 5 // CHECK4-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 7
// CHECK4-NEXT: store i8* null, i8** [[TMP130]], align 4 // CHECK4-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to double**
// CHECK4-NEXT: [[TMP131:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 6 // CHECK4-NEXT: store double* [[VLA1]], double** [[TMP131]], align 4
// CHECK4-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to i32* // CHECK4-NEXT: [[TMP132:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 7
// CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP132]], align 4 // CHECK4-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to double**
// CHECK4-NEXT: [[TMP133:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 6 // CHECK4-NEXT: store double* [[VLA1]], double** [[TMP133]], align 4
// CHECK4-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to i32* // CHECK4-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 7), align 4
// CHECK4-NEXT: store i32 [[TMP3]], i32* [[TMP134]], align 4 // CHECK4-NEXT: [[TMP134:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 7
// CHECK4-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK4-NEXT: store i8* null, i8** [[TMP134]], align 4
// CHECK4-NEXT: store i64 4, i64* [[TMP135]], align 4 // CHECK4-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 8
// CHECK4-NEXT: [[TMP136:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 6 // CHECK4-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT**
// CHECK4-NEXT: store i8* null, i8** [[TMP136]], align 4 // CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 4
// CHECK4-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 7 // CHECK4-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 8
// CHECK4-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to double** // CHECK4-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT**
// CHECK4-NEXT: store double* [[VLA1]], double** [[TMP138]], align 4 // CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 4
// CHECK4-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 7 // CHECK4-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 8
// CHECK4-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to double** // CHECK4-NEXT: store i8* null, i8** [[TMP139]], align 4
// CHECK4-NEXT: store double* [[VLA1]], double** [[TMP140]], align 4 // CHECK4-NEXT: [[TMP140:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK4-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK4-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK4-NEXT: store i64 [[TMP94]], i64* [[TMP141]], align 4
// CHECK4-NEXT: [[TMP142:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 7
// CHECK4-NEXT: store i8* null, i8** [[TMP142]], align 4
// CHECK4-NEXT: [[TMP143:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 8
// CHECK4-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to %struct.TT**
// CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP144]], align 4
// CHECK4-NEXT: [[TMP145:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 8
// CHECK4-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to %struct.TT**
// CHECK4-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP146]], align 4
// CHECK4-NEXT: [[TMP147:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK4-NEXT: store i64 12, i64* [[TMP147]], align 4
// CHECK4-NEXT: [[TMP148:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 8
// CHECK4-NEXT: store i8* null, i8** [[TMP148]], align 4
// CHECK4-NEXT: [[TMP149:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK4-NEXT: [[TMP150:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK4-NEXT: [[TMP151:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK4-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK4-NEXT: [[TMP142:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP140]], i8** [[TMP141]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK4-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0 // CHECK4-NEXT: [[TMP143:%.*]] = icmp ne i32 [[TMP142]], 0
// CHECK4-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED25:%.*]], label [[OMP_OFFLOAD_CONT26:%.*]] // CHECK4-NEXT: br i1 [[TMP143]], label [[OMP_OFFLOAD_FAILED25:%.*]], label [[OMP_OFFLOAD_CONT26:%.*]]
// CHECK4: omp_offload.failed25: // CHECK4: omp_offload.failed25:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT26]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT26]]
// CHECK4: omp_offload.cont26: // CHECK4: omp_offload.cont26:
// CHECK4-NEXT: br label [[OMP_IF_END28:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END28:%.*]]
// CHECK4: omp_if.else27: // CHECK4: omp_if.else27:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_IF_END28]] // CHECK4-NEXT: br label [[OMP_IF_END28]]
// CHECK4: omp_if.end28: // CHECK4: omp_if.end28:
// CHECK4-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK4-NEXT: [[TMP144:%.*]] = load i32, i32* [[A]], align 4
// CHECK4-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: [[TMP145:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP145]])
// CHECK4-NEXT: ret i32 [[TMP154]] // CHECK4-NEXT: ret i32 [[TMP144]]
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97
// CHECK4-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK4-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 672 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK4-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK4-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK4-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK4-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK4-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK4-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK4-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 14 Lines
// CHECK4-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2 // CHECK4-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP7]], 2
// CHECK4-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64 // CHECK4-NEXT: [[TMP9:%.*]] = sext i32 [[TMP8]] to i64
// CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1** // CHECK4-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %struct.S1**
// CHECK4-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4 // CHECK4-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP11]], align 4
// CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double** // CHECK4-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to double**
// CHECK4-NEXT: store double* [[A]], double** [[TMP13]], align 4 // CHECK4-NEXT: store double* [[A]], double** [[TMP13]], align 4
// CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK4-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK4-NEXT: store i64 8, i64* [[TMP14]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP14]], align 4
// CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK4-NEXT: store i8* null, i8** [[TMP15]], align 4 // CHECK4-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK4-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK4-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32*
// CHECK4-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP18]], align 4
// CHECK4-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32* // CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[TMP19]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK4-NEXT: store i64 4, i64* [[TMP20]], align 4 // CHECK4-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK4-NEXT: store i32 2, i32* [[TMP21]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP21]], align 4 // CHECK4-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK4-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32*
// CHECK4-NEXT: store i32 2, i32* [[TMP23]], align 4 // CHECK4-NEXT: store i32 2, i32* [[TMP23]], align 4
// CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32* // CHECK4-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK4-NEXT: store i32 2, i32* [[TMP25]], align 4 // CHECK4-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK4-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK4-NEXT: store i64 4, i64* [[TMP26]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP26]], align 4
// CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK4-NEXT: store i8* null, i8** [[TMP27]], align 4 // CHECK4-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP28]], align 4
// CHECK4-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK4-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK4-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK4-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK4-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32* // CHECK4-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i16**
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4 // CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP31]], align 4
// CHECK4-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK4-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK4-NEXT: store i64 4, i64* [[TMP32]], align 4 // CHECK4-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i16**
// CHECK4-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP33]], align 4
// CHECK4-NEXT: store i8* null, i8** [[TMP33]], align 4 // CHECK4-NEXT: store i64 [[TMP9]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 4), align 4
// CHECK4-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK4-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK4-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i16** // CHECK4-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP35]], align 4 // CHECK4-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK4-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i16**
// CHECK4-NEXT: store i16* [[VLA]], i16** [[TMP37]], align 4
// CHECK4-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK4-NEXT: store i64 [[TMP9]], i64* [[TMP38]], align 4
// CHECK4-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK4-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK4-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK4-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 5, i8** [[TMP40]], i8** [[TMP41]], i64* [[TMP42]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK4-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 5, i8** [[TMP35]], i8** [[TMP36]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK4-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0 // CHECK4-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK4-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP5]], i32 2, i32 [[TMP1]], i16* [[VLA]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_IF_END]] // CHECK4-NEXT: br label [[OMP_IF_END]]
// CHECK4: omp_if.end: // CHECK4: omp_if.end:
// CHECK4-NEXT: [[TMP45:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK4-NEXT: [[TMP39:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP45]] // CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP39]]
// CHECK4-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK4-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK4-NEXT: [[TMP46:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2 // CHECK4-NEXT: [[TMP40:%.*]] = load i16, i16* [[ARRAYIDX2]], align 2
// CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP46]] to i32 // CHECK4-NEXT: [[CONV:%.*]] = sext i16 [[TMP40]] to i32
// CHECK4-NEXT: [[TMP47:%.*]] = load i32, i32* [[B]], align 4 // CHECK4-NEXT: [[TMP41:%.*]] = load i32, i32* [[B]], align 4
// CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP47]] // CHECK4-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[TMP41]]
// CHECK4-NEXT: [[TMP48:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK4-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP48]]) // CHECK4-NEXT: call void @llvm.stackrestore(i8* [[TMP42]])
// CHECK4-NEXT: ret i32 [[ADD3]] // CHECK4-NEXT: ret i32 [[ADD3]]
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK4-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK4-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK4-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1 // CHECK4-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1
// CHECK4-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK4-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK4-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1 // CHECK4-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1
// CHECK4-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK4-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK4-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK4-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK4-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1 // CHECK4-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1
// CHECK4-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64 // CHECK4-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64
// CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK4-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK4-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK4-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK4-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK4-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
▲ Show 20 Lines • Show All 53 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK4-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK4-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK4-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK4-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK4-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK4-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK4-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK4-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK4-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK4-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK4-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK4: omp_offload.failed: // CHECK4: omp_offload.failed:
// CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK4-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK4-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK4: omp_offload.cont: // CHECK4: omp_offload.cont:
// CHECK4-NEXT: br label [[OMP_IF_END:%.*]] // CHECK4-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK4: omp_if.else: // CHECK4: omp_if.else:
Show All 20 Lines
// CHECK4-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4 // CHECK4-NEXT: store i16* [[C]], i16** [[C_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[VLA_ADDR2]], align 4
// CHECK4-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4 // CHECK4-NEXT: [[TMP3:%.*]] = load i16*, i16** [[C_ADDR]], align 4
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK4-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 112 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK4-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK4-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK4-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK4-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2 // CHECK4-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2
// CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK4-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK4-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK4-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK4-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK4-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1 // CHECK4-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1
// CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[N:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[N:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 141 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK4-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK4-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK4-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK4-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK4-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK4-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK4-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
// CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK4-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 114 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [2 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [2 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [2 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [2 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8
// CHECK5-NEXT: [[_TMP19:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[_TMP19:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[A_CASTED22:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[A_CASTED22:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS26:%.*]] = alloca [9 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS26:%.*]] = alloca [9 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_PTRS27:%.*]] = alloca [9 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS27:%.*]] = alloca [9 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS28:%.*]] = alloca [9 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS28:%.*]] = alloca [9 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 8
// CHECK5-NEXT: [[_TMP29:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[_TMP29:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK5-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK5-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK5-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK5-NEXT: store i32 0, i32* [[A]], align 4 // CHECK5-NEXT: store i32 0, i32* [[A]], align 4
// CHECK5-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK5-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK5-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK5-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK5-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
▲ Show 20 Lines • Show All 162 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[TMP93:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK5-NEXT: [[TMP93:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK5-NEXT: [[TMP94:%.*]] = mul nuw i64 [[TMP93]], 8 // CHECK5-NEXT: [[TMP94:%.*]] = mul nuw i64 [[TMP93]], 8
// CHECK5-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0 // CHECK5-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK5-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64* // CHECK5-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64*
// CHECK5-NEXT: store i64 [[TMP90]], i64* [[TMP96]], align 8 // CHECK5-NEXT: store i64 [[TMP90]], i64* [[TMP96]], align 8
// CHECK5-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0 // CHECK5-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK5-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i64* // CHECK5-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i64*
// CHECK5-NEXT: store i64 [[TMP90]], i64* [[TMP98]], align 8 // CHECK5-NEXT: store i64 [[TMP90]], i64* [[TMP98]], align 8
// CHECK5-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK5-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 0
// CHECK5-NEXT: store i64 4, i64* [[TMP99]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP99]], align 8
// CHECK5-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 0 // CHECK5-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 1
// CHECK5-NEXT: store i8* null, i8** [[TMP100]], align 8 // CHECK5-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to [10 x float]**
// CHECK5-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 1 // CHECK5-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP101]], align 8
// CHECK5-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to [10 x float]** // CHECK5-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 1
// CHECK5-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP102]], align 8 // CHECK5-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to [10 x float]**
// CHECK5-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 1 // CHECK5-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP103]], align 8
// CHECK5-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]** // CHECK5-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 1
// CHECK5-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP104]], align 8
// CHECK5-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK5-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 2
// CHECK5-NEXT: store i64 40, i64* [[TMP105]], align 8 // CHECK5-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to i64*
// CHECK5-NEXT: [[TMP106:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 1 // CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP106]], align 8
// CHECK5-NEXT: store i8* null, i8** [[TMP106]], align 8 // CHECK5-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 2
// CHECK5-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 2
// CHECK5-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i64* // CHECK5-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i64*
// CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP108]], align 8 // CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP108]], align 8
// CHECK5-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 2 // CHECK5-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 2
// CHECK5-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to i64* // CHECK5-NEXT: store i8* null, i8** [[TMP109]], align 8
// CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP110]], align 8 // CHECK5-NEXT: [[TMP110:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 3
// CHECK5-NEXT: [[TMP111:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK5-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to float**
// CHECK5-NEXT: store i64 8, i64* [[TMP111]], align 8 // CHECK5-NEXT: store float* [[VLA]], float** [[TMP111]], align 8
// CHECK5-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 2 // CHECK5-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 3
// CHECK5-NEXT: store i8* null, i8** [[TMP112]], align 8 // CHECK5-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to float**
// CHECK5-NEXT: [[TMP113:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 3 // CHECK5-NEXT: store float* [[VLA]], float** [[TMP113]], align 8
// CHECK5-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float** // CHECK5-NEXT: store i64 [[TMP92]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 3), align 8
// CHECK5-NEXT: store float* [[VLA]], float** [[TMP114]], align 8 // CHECK5-NEXT: [[TMP114:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 3
// CHECK5-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 3 // CHECK5-NEXT: store i8* null, i8** [[TMP114]], align 8
// CHECK5-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float** // CHECK5-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 4
// CHECK5-NEXT: store float* [[VLA]], float** [[TMP116]], align 8 // CHECK5-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]**
// CHECK5-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK5-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 8
// CHECK5-NEXT: store i64 [[TMP92]], i64* [[TMP117]], align 8 // CHECK5-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 4
// CHECK5-NEXT: [[TMP118:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 3 // CHECK5-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to [5 x [10 x double]]**
// CHECK5-NEXT: store i8* null, i8** [[TMP118]], align 8 // CHECK5-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP118]], align 8
// CHECK5-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 4 // CHECK5-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 4
// CHECK5-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to [5 x [10 x double]]** // CHECK5-NEXT: store i8* null, i8** [[TMP119]], align 8
// CHECK5-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP120]], align 8 // CHECK5-NEXT: [[TMP120:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 5
// CHECK5-NEXT: [[TMP121:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 4 // CHECK5-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i64*
// CHECK5-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to [5 x [10 x double]]** // CHECK5-NEXT: store i64 5, i64* [[TMP121]], align 8
// CHECK5-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP122]], align 8 // CHECK5-NEXT: [[TMP122:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 5
// CHECK5-NEXT: [[TMP123:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK5-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i64*
// CHECK5-NEXT: store i64 400, i64* [[TMP123]], align 8 // CHECK5-NEXT: store i64 5, i64* [[TMP123]], align 8
// CHECK5-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 4 // CHECK5-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 5
// CHECK5-NEXT: store i8* null, i8** [[TMP124]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP124]], align 8
// CHECK5-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 5 // CHECK5-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 6
// CHECK5-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64* // CHECK5-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64*
// CHECK5-NEXT: store i64 5, i64* [[TMP126]], align 8 // CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP126]], align 8
// CHECK5-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 5 // CHECK5-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 6
// CHECK5-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64* // CHECK5-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64*
// CHECK5-NEXT: store i64 5, i64* [[TMP128]], align 8 // CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP128]], align 8
// CHECK5-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK5-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 6
// CHECK5-NEXT: store i64 8, i64* [[TMP129]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP129]], align 8
// CHECK5-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 5 // CHECK5-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 7
// CHECK5-NEXT: store i8* null, i8** [[TMP130]], align 8 // CHECK5-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to double**
// CHECK5-NEXT: [[TMP131:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 6 // CHECK5-NEXT: store double* [[VLA1]], double** [[TMP131]], align 8
// CHECK5-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to i64* // CHECK5-NEXT: [[TMP132:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 7
// CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP132]], align 8 // CHECK5-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to double**
// CHECK5-NEXT: [[TMP133:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 6 // CHECK5-NEXT: store double* [[VLA1]], double** [[TMP133]], align 8
// CHECK5-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to i64* // CHECK5-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 7), align 8
// CHECK5-NEXT: store i64 [[TMP5]], i64* [[TMP134]], align 8 // CHECK5-NEXT: [[TMP134:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 7
// CHECK5-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK5-NEXT: store i8* null, i8** [[TMP134]], align 8
// CHECK5-NEXT: store i64 8, i64* [[TMP135]], align 8 // CHECK5-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 8
// CHECK5-NEXT: [[TMP136:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 6 // CHECK5-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT**
// CHECK5-NEXT: store i8* null, i8** [[TMP136]], align 8 // CHECK5-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 8
// CHECK5-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 7 // CHECK5-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 8
// CHECK5-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to double** // CHECK5-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT**
// CHECK5-NEXT: store double* [[VLA1]], double** [[TMP138]], align 8 // CHECK5-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 8
// CHECK5-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 7 // CHECK5-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 8
// CHECK5-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to double** // CHECK5-NEXT: store i8* null, i8** [[TMP139]], align 8
// CHECK5-NEXT: store double* [[VLA1]], double** [[TMP140]], align 8 // CHECK5-NEXT: [[TMP140:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK5-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK5-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK5-NEXT: store i64 [[TMP94]], i64* [[TMP141]], align 8
// CHECK5-NEXT: [[TMP142:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 7
// CHECK5-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK5-NEXT: [[TMP143:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 8
// CHECK5-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to %struct.TT**
// CHECK5-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP144]], align 8
// CHECK5-NEXT: [[TMP145:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 8
// CHECK5-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to %struct.TT**
// CHECK5-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP146]], align 8
// CHECK5-NEXT: [[TMP147:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK5-NEXT: store i64 16, i64* [[TMP147]], align 8
// CHECK5-NEXT: [[TMP148:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 8
// CHECK5-NEXT: store i8* null, i8** [[TMP148]], align 8
// CHECK5-NEXT: [[TMP149:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK5-NEXT: [[TMP150:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK5-NEXT: [[TMP151:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK5-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK5-NEXT: [[TMP142:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP140]], i8** [[TMP141]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK5-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0 // CHECK5-NEXT: [[TMP143:%.*]] = icmp ne i32 [[TMP142]], 0
// CHECK5-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]] // CHECK5-NEXT: br i1 [[TMP143]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK5: omp_offload.failed30: // CHECK5: omp_offload.failed30:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT31]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK5: omp_offload.cont31: // CHECK5: omp_offload.cont31:
// CHECK5-NEXT: br label [[OMP_IF_END33:%.*]] // CHECK5-NEXT: br label [[OMP_IF_END33:%.*]]
// CHECK5: omp_if.else32: // CHECK5: omp_if.else32:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_IF_END33]] // CHECK5-NEXT: br label [[OMP_IF_END33]]
// CHECK5: omp_if.end33: // CHECK5: omp_if.end33:
// CHECK5-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK5-NEXT: [[TMP144:%.*]] = load i32, i32* [[A]], align 4
// CHECK5-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK5-NEXT: [[TMP145:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK5-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK5-NEXT: call void @llvm.stackrestore(i8* [[TMP145]])
// CHECK5-NEXT: ret i32 [[TMP154]] // CHECK5-NEXT: ret i32 [[TMP144]]
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97
// CHECK5-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK5-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 687 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK5-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK5-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK5-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 8 // CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [6 x i64], align 8
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK5-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK5-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK5-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK5-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK5-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK5-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK5-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK5-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 25 Lines
// CHECK5-NEXT: [[TMP11:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK5-NEXT: [[TMP11:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK5-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], 2 // CHECK5-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], 2
// CHECK5-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1** // CHECK5-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1**
// CHECK5-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 8 // CHECK5-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 8
// CHECK5-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double** // CHECK5-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double**
// CHECK5-NEXT: store double* [[A]], double** [[TMP16]], align 8 // CHECK5-NEXT: store double* [[A]], double** [[TMP16]], align 8
// CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK5-NEXT: store i64 8, i64* [[TMP17]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP17]], align 8
// CHECK5-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK5-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK5-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK5-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64*
// CHECK5-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK5-NEXT: store i64 [[TMP7]], i64* [[TMP19]], align 8
// CHECK5-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64* // CHECK5-NEXT: [[TMP20:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK5-NEXT: store i64 [[TMP7]], i64* [[TMP20]], align 8 // CHECK5-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK5-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK5-NEXT: store i64 [[TMP7]], i64* [[TMP21]], align 8
// CHECK5-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i64* // CHECK5-NEXT: [[TMP22:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK5-NEXT: store i64 [[TMP7]], i64* [[TMP22]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK5-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK5-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK5-NEXT: store i64 4, i64* [[TMP23]], align 8 // CHECK5-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i64*
// CHECK5-NEXT: [[TMP24:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK5-NEXT: store i64 2, i64* [[TMP24]], align 8
// CHECK5-NEXT: store i8* null, i8** [[TMP24]], align 8 // CHECK5-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK5-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK5-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64* // CHECK5-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK5-NEXT: store i64 2, i64* [[TMP26]], align 8 // CHECK5-NEXT: store i64 2, i64* [[TMP26]], align 8
// CHECK5-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK5-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK5-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64* // CHECK5-NEXT: store i8* null, i8** [[TMP27]], align 8
// CHECK5-NEXT: store i64 2, i64* [[TMP28]], align 8 // CHECK5-NEXT: [[TMP28:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK5-NEXT: [[TMP29:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK5-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64*
// CHECK5-NEXT: store i64 8, i64* [[TMP29]], align 8 // CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8
// CHECK5-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK5-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK5-NEXT: store i8* null, i8** [[TMP30]], align 8 // CHECK5-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64*
// CHECK5-NEXT: [[TMP31:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8
// CHECK5-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i64* // CHECK5-NEXT: [[TMP32:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP32]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP32]], align 8
// CHECK5-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK5-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK5-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i64* // CHECK5-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i16**
// CHECK5-NEXT: store i64 [[TMP2]], i64* [[TMP34]], align 8 // CHECK5-NEXT: store i16* [[VLA]], i16** [[TMP34]], align 8
// CHECK5-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK5-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK5-NEXT: store i64 8, i64* [[TMP35]], align 8 // CHECK5-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i16**
// CHECK5-NEXT: [[TMP36:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK5-NEXT: store i16* [[VLA]], i16** [[TMP36]], align 8
// CHECK5-NEXT: store i8* null, i8** [[TMP36]], align 8 // CHECK5-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.13, i32 0, i32 4), align 8
// CHECK5-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK5-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK5-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i16** // CHECK5-NEXT: store i8* null, i8** [[TMP37]], align 8
// CHECK5-NEXT: store i16* [[VLA]], i16** [[TMP38]], align 8 // CHECK5-NEXT: [[TMP38:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK5-NEXT: [[TMP39:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK5-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64*
// CHECK5-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i16** // CHECK5-NEXT: store i64 [[TMP9]], i64* [[TMP39]], align 8
// CHECK5-NEXT: store i16* [[VLA]], i16** [[TMP40]], align 8 // CHECK5-NEXT: [[TMP40:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK5-NEXT: [[TMP41:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK5-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i64*
// CHECK5-NEXT: store i64 [[TMP12]], i64* [[TMP41]], align 8 // CHECK5-NEXT: store i64 [[TMP9]], i64* [[TMP41]], align 8
// CHECK5-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK5-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK5-NEXT: store i8* null, i8** [[TMP42]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP42]], align 8
// CHECK5-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK5-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i64* // CHECK5-NEXT: [[TMP44:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: store i64 [[TMP9]], i64* [[TMP44]], align 8
// CHECK5-NEXT: [[TMP45:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK5-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i64*
// CHECK5-NEXT: store i64 [[TMP9]], i64* [[TMP46]], align 8
// CHECK5-NEXT: [[TMP47:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5
// CHECK5-NEXT: store i64 1, i64* [[TMP47]], align 8
// CHECK5-NEXT: [[TMP48:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK5-NEXT: store i8* null, i8** [[TMP48]], align 8
// CHECK5-NEXT: [[TMP49:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP50:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP51:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK5-NEXT: [[TMP52:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 6, i8** [[TMP49]], i8** [[TMP50]], i64* [[TMP51]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK5-NEXT: [[TMP45:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 6, i8** [[TMP43]], i8** [[TMP44]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK5-NEXT: [[TMP53:%.*]] = icmp ne i32 [[TMP52]], 0 // CHECK5-NEXT: [[TMP46:%.*]] = icmp ne i32 [[TMP45]], 0
// CHECK5-NEXT: br i1 [[TMP53]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK5-NEXT: br i1 [[TMP46]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK5: omp_offload.failed: // CHECK5: omp_offload.failed:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK5: omp_offload.cont: // CHECK5: omp_offload.cont:
// CHECK5-NEXT: br label [[OMP_IF_END:%.*]] // CHECK5-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK5: omp_if.else: // CHECK5: omp_if.else:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_IF_END]] // CHECK5-NEXT: br label [[OMP_IF_END]]
// CHECK5: omp_if.end: // CHECK5: omp_if.end:
// CHECK5-NEXT: [[TMP54:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK5-NEXT: [[TMP47:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP54]] // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP47]]
// CHECK5-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK5-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK5-NEXT: [[TMP55:%.*]] = load i16, i16* [[ARRAYIDX5]], align 2 // CHECK5-NEXT: [[TMP48:%.*]] = load i16, i16* [[ARRAYIDX5]], align 2
// CHECK5-NEXT: [[CONV6:%.*]] = sext i16 [[TMP55]] to i32 // CHECK5-NEXT: [[CONV6:%.*]] = sext i16 [[TMP48]] to i32
// CHECK5-NEXT: [[TMP56:%.*]] = load i32, i32* [[B]], align 4 // CHECK5-NEXT: [[TMP49:%.*]] = load i32, i32* [[B]], align 4
// CHECK5-NEXT: [[ADD7:%.*]] = add nsw i32 [[CONV6]], [[TMP56]] // CHECK5-NEXT: [[ADD7:%.*]] = add nsw i32 [[CONV6]], [[TMP49]]
// CHECK5-NEXT: [[TMP57:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK5-NEXT: [[TMP50:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK5-NEXT: call void @llvm.stackrestore(i8* [[TMP57]]) // CHECK5-NEXT: call void @llvm.stackrestore(i8* [[TMP50]])
// CHECK5-NEXT: ret i32 [[ADD7]] // CHECK5-NEXT: ret i32 [[ADD7]]
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK5-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK5-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK5-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1
// CHECK5-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK5-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK5-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1 // CHECK5-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1
// CHECK5-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK5-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK5-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK5-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK5-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1 // CHECK5-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1
// CHECK5-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64 // CHECK5-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64
// CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK5-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK5-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK5-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK5-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK5-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK5-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK5: omp_offload.failed: // CHECK5: omp_offload.failed:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK5: omp_offload.cont: // CHECK5: omp_offload.cont:
// CHECK5-NEXT: br label [[OMP_IF_END:%.*]] // CHECK5-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK5: omp_if.else: // CHECK5: omp_if.else:
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK5-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK5-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK5-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK5-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK5-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK5-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK5-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK5-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK5-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK5-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK5-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK5-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK5-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK5-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK5-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK5: omp_offload.failed: // CHECK5: omp_offload.failed:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK5: omp_offload.cont: // CHECK5: omp_offload.cont:
// CHECK5-NEXT: br label [[OMP_IF_END:%.*]] // CHECK5-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK5: omp_if.else: // CHECK5: omp_if.else:
Show All 32 Lines
// CHECK5-NEXT: store i32 [[TMP4]], i32* [[CONV4]], align 4 // CHECK5-NEXT: store i32 [[TMP4]], i32* [[CONV4]], align 4
// CHECK5-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK5-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK5-NEXT: [[TMP6:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK5-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP6]] to i1 // CHECK5-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP6]] to i1
// CHECK5-NEXT: [[CONV5:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8* // CHECK5-NEXT: [[CONV5:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
// CHECK5-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 // CHECK5-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
// CHECK5-NEXT: store i8 [[FROMBOOL]], i8* [[CONV5]], align 1 // CHECK5-NEXT: store i8 [[FROMBOOL]], i8* [[CONV5]], align 1
// CHECK5-NEXT: [[TMP7:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK5-NEXT: [[TMP7:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]], i64 [[TMP7]]) // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*, i64)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]], i64 [[TMP7]])
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK5-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 162 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2 // CHECK5-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2
// CHECK5-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK5-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK5-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2 // CHECK5-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2
// CHECK5-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK5-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK5-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK5-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK5-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK5-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK5-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1 // CHECK5-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1
// CHECK5-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK5-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[N:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[N:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 145 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK5-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK5-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK5-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK5-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK5-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK5-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK5-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK5-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK5-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK5-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK5-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK5-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK5-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [2 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS16:%.*]] = alloca [2 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [2 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_PTRS17:%.*]] = alloca [2 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS18:%.*]] = alloca [2 x i8*], align 8
// CHECK6-NEXT: [[_TMP19:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[_TMP19:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[A_CASTED22:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[A_CASTED22:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS26:%.*]] = alloca [9 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS26:%.*]] = alloca [9 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_PTRS27:%.*]] = alloca [9 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_PTRS27:%.*]] = alloca [9 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS28:%.*]] = alloca [9 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS28:%.*]] = alloca [9 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 8
// CHECK6-NEXT: [[_TMP29:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[_TMP29:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK6-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK6-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK6-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK6-NEXT: store i32 0, i32* [[A]], align 4 // CHECK6-NEXT: store i32 0, i32* [[A]], align 4
// CHECK6-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK6-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64 // CHECK6-NEXT: [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK6-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK6-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
▲ Show 20 Lines • Show All 162 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[TMP93:%.*]] = mul nuw i64 5, [[TMP5]] // CHECK6-NEXT: [[TMP93:%.*]] = mul nuw i64 5, [[TMP5]]
// CHECK6-NEXT: [[TMP94:%.*]] = mul nuw i64 [[TMP93]], 8 // CHECK6-NEXT: [[TMP94:%.*]] = mul nuw i64 [[TMP93]], 8
// CHECK6-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0 // CHECK6-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK6-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64* // CHECK6-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i64*
// CHECK6-NEXT: store i64 [[TMP90]], i64* [[TMP96]], align 8 // CHECK6-NEXT: store i64 [[TMP90]], i64* [[TMP96]], align 8
// CHECK6-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0 // CHECK6-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK6-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i64* // CHECK6-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i64*
// CHECK6-NEXT: store i64 [[TMP90]], i64* [[TMP98]], align 8 // CHECK6-NEXT: store i64 [[TMP90]], i64* [[TMP98]], align 8
// CHECK6-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK6-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 0
// CHECK6-NEXT: store i64 4, i64* [[TMP99]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP99]], align 8
// CHECK6-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 0 // CHECK6-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 1
// CHECK6-NEXT: store i8* null, i8** [[TMP100]], align 8 // CHECK6-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to [10 x float]**
// CHECK6-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 1 // CHECK6-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP101]], align 8
// CHECK6-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to [10 x float]** // CHECK6-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 1
// CHECK6-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP102]], align 8 // CHECK6-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to [10 x float]**
// CHECK6-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 1 // CHECK6-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP103]], align 8
// CHECK6-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]** // CHECK6-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 1
// CHECK6-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP104]], align 8
// CHECK6-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK6-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 2
// CHECK6-NEXT: store i64 40, i64* [[TMP105]], align 8 // CHECK6-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to i64*
// CHECK6-NEXT: [[TMP106:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 1 // CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP106]], align 8
// CHECK6-NEXT: store i8* null, i8** [[TMP106]], align 8 // CHECK6-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 2
// CHECK6-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 2
// CHECK6-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i64* // CHECK6-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i64*
// CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP108]], align 8 // CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP108]], align 8
// CHECK6-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 2 // CHECK6-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 2
// CHECK6-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to i64* // CHECK6-NEXT: store i8* null, i8** [[TMP109]], align 8
// CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP110]], align 8 // CHECK6-NEXT: [[TMP110:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 3
// CHECK6-NEXT: [[TMP111:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK6-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to float**
// CHECK6-NEXT: store i64 8, i64* [[TMP111]], align 8 // CHECK6-NEXT: store float* [[VLA]], float** [[TMP111]], align 8
// CHECK6-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 2 // CHECK6-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 3
// CHECK6-NEXT: store i8* null, i8** [[TMP112]], align 8 // CHECK6-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to float**
// CHECK6-NEXT: [[TMP113:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 3 // CHECK6-NEXT: store float* [[VLA]], float** [[TMP113]], align 8
// CHECK6-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float** // CHECK6-NEXT: store i64 [[TMP92]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 3), align 8
// CHECK6-NEXT: store float* [[VLA]], float** [[TMP114]], align 8 // CHECK6-NEXT: [[TMP114:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 3
// CHECK6-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 3 // CHECK6-NEXT: store i8* null, i8** [[TMP114]], align 8
// CHECK6-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float** // CHECK6-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 4
// CHECK6-NEXT: store float* [[VLA]], float** [[TMP116]], align 8 // CHECK6-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]**
// CHECK6-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK6-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 8
// CHECK6-NEXT: store i64 [[TMP92]], i64* [[TMP117]], align 8 // CHECK6-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 4
// CHECK6-NEXT: [[TMP118:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 3 // CHECK6-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to [5 x [10 x double]]**
// CHECK6-NEXT: store i8* null, i8** [[TMP118]], align 8 // CHECK6-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP118]], align 8
// CHECK6-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 4 // CHECK6-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 4
// CHECK6-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to [5 x [10 x double]]** // CHECK6-NEXT: store i8* null, i8** [[TMP119]], align 8
// CHECK6-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP120]], align 8 // CHECK6-NEXT: [[TMP120:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 5
// CHECK6-NEXT: [[TMP121:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 4 // CHECK6-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i64*
// CHECK6-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to [5 x [10 x double]]** // CHECK6-NEXT: store i64 5, i64* [[TMP121]], align 8
// CHECK6-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP122]], align 8 // CHECK6-NEXT: [[TMP122:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 5
// CHECK6-NEXT: [[TMP123:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK6-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i64*
// CHECK6-NEXT: store i64 400, i64* [[TMP123]], align 8 // CHECK6-NEXT: store i64 5, i64* [[TMP123]], align 8
// CHECK6-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 4 // CHECK6-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 5
// CHECK6-NEXT: store i8* null, i8** [[TMP124]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP124]], align 8
// CHECK6-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 5 // CHECK6-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 6
// CHECK6-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64* // CHECK6-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i64*
// CHECK6-NEXT: store i64 5, i64* [[TMP126]], align 8 // CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP126]], align 8
// CHECK6-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 5 // CHECK6-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 6
// CHECK6-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64* // CHECK6-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i64*
// CHECK6-NEXT: store i64 5, i64* [[TMP128]], align 8 // CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP128]], align 8
// CHECK6-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK6-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 6
// CHECK6-NEXT: store i64 8, i64* [[TMP129]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP129]], align 8
// CHECK6-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 5 // CHECK6-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 7
// CHECK6-NEXT: store i8* null, i8** [[TMP130]], align 8 // CHECK6-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to double**
// CHECK6-NEXT: [[TMP131:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 6 // CHECK6-NEXT: store double* [[VLA1]], double** [[TMP131]], align 8
// CHECK6-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to i64* // CHECK6-NEXT: [[TMP132:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 7
// CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP132]], align 8 // CHECK6-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to double**
// CHECK6-NEXT: [[TMP133:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 6 // CHECK6-NEXT: store double* [[VLA1]], double** [[TMP133]], align 8
// CHECK6-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to i64* // CHECK6-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 7), align 8
// CHECK6-NEXT: store i64 [[TMP5]], i64* [[TMP134]], align 8 // CHECK6-NEXT: [[TMP134:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 7
// CHECK6-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK6-NEXT: store i8* null, i8** [[TMP134]], align 8
// CHECK6-NEXT: store i64 8, i64* [[TMP135]], align 8 // CHECK6-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 8
// CHECK6-NEXT: [[TMP136:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 6 // CHECK6-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT**
// CHECK6-NEXT: store i8* null, i8** [[TMP136]], align 8 // CHECK6-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 8
// CHECK6-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 7 // CHECK6-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 8
// CHECK6-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to double** // CHECK6-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT**
// CHECK6-NEXT: store double* [[VLA1]], double** [[TMP138]], align 8 // CHECK6-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 8
// CHECK6-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 7 // CHECK6-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 8
// CHECK6-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to double** // CHECK6-NEXT: store i8* null, i8** [[TMP139]], align 8
// CHECK6-NEXT: store double* [[VLA1]], double** [[TMP140]], align 8 // CHECK6-NEXT: [[TMP140:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK6-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK6-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK6-NEXT: store i64 [[TMP94]], i64* [[TMP141]], align 8
// CHECK6-NEXT: [[TMP142:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 7
// CHECK6-NEXT: store i8* null, i8** [[TMP142]], align 8
// CHECK6-NEXT: [[TMP143:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 8
// CHECK6-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to %struct.TT**
// CHECK6-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP144]], align 8
// CHECK6-NEXT: [[TMP145:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 8
// CHECK6-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to %struct.TT**
// CHECK6-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP146]], align 8
// CHECK6-NEXT: [[TMP147:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK6-NEXT: store i64 16, i64* [[TMP147]], align 8
// CHECK6-NEXT: [[TMP148:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 8
// CHECK6-NEXT: store i8* null, i8** [[TMP148]], align 8
// CHECK6-NEXT: [[TMP149:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK6-NEXT: [[TMP150:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK6-NEXT: [[TMP151:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK6-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK6-NEXT: [[TMP142:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP140]], i8** [[TMP141]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK6-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0 // CHECK6-NEXT: [[TMP143:%.*]] = icmp ne i32 [[TMP142]], 0
// CHECK6-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]] // CHECK6-NEXT: br i1 [[TMP143]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK6: omp_offload.failed30: // CHECK6: omp_offload.failed30:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT31]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK6: omp_offload.cont31: // CHECK6: omp_offload.cont31:
// CHECK6-NEXT: br label [[OMP_IF_END33:%.*]] // CHECK6-NEXT: br label [[OMP_IF_END33:%.*]]
// CHECK6: omp_if.else32: // CHECK6: omp_if.else32:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i64 [[TMP90]], [10 x float]* [[B]], i64 [[TMP2]], float* [[VLA]], [5 x [10 x double]]* [[C]], i64 5, i64 [[TMP5]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_IF_END33]] // CHECK6-NEXT: br label [[OMP_IF_END33]]
// CHECK6: omp_if.end33: // CHECK6: omp_if.end33:
// CHECK6-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK6-NEXT: [[TMP144:%.*]] = load i32, i32* [[A]], align 4
// CHECK6-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK6-NEXT: [[TMP145:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK6-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK6-NEXT: call void @llvm.stackrestore(i8* [[TMP145]])
// CHECK6-NEXT: ret i32 [[TMP154]] // CHECK6-NEXT: ret i32 [[TMP144]]
// //
// //
// CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97 // CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97
// CHECK6-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK6-SAME: (i64 [[AA:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]], i64 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK6-NEXT: entry: // CHECK6-NEXT: entry:
// CHECK6-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 687 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK6-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK6-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK6-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK6-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 8 // CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [6 x i64], align 8
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8 // CHECK6-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 8
// CHECK6-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK6-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK6-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8 // CHECK6-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 8
// CHECK6-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK6-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK6-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK6-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 25 Lines
// CHECK6-NEXT: [[TMP11:%.*]] = mul nuw i64 2, [[TMP2]] // CHECK6-NEXT: [[TMP11:%.*]] = mul nuw i64 2, [[TMP2]]
// CHECK6-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], 2 // CHECK6-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], 2
// CHECK6-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1** // CHECK6-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1**
// CHECK6-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 8 // CHECK6-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 8
// CHECK6-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double** // CHECK6-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double**
// CHECK6-NEXT: store double* [[A]], double** [[TMP16]], align 8 // CHECK6-NEXT: store double* [[A]], double** [[TMP16]], align 8
// CHECK6-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK6-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK6-NEXT: store i64 8, i64* [[TMP17]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP17]], align 8
// CHECK6-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK6-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK6-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK6-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64*
// CHECK6-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK6-NEXT: store i64 [[TMP7]], i64* [[TMP19]], align 8
// CHECK6-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64* // CHECK6-NEXT: [[TMP20:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK6-NEXT: store i64 [[TMP7]], i64* [[TMP20]], align 8 // CHECK6-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK6-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK6-NEXT: store i64 [[TMP7]], i64* [[TMP21]], align 8
// CHECK6-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i64* // CHECK6-NEXT: [[TMP22:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK6-NEXT: store i64 [[TMP7]], i64* [[TMP22]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK6-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK6-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK6-NEXT: store i64 4, i64* [[TMP23]], align 8 // CHECK6-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i64*
// CHECK6-NEXT: [[TMP24:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK6-NEXT: store i64 2, i64* [[TMP24]], align 8
// CHECK6-NEXT: store i8* null, i8** [[TMP24]], align 8 // CHECK6-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK6-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK6-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64* // CHECK6-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK6-NEXT: store i64 2, i64* [[TMP26]], align 8 // CHECK6-NEXT: store i64 2, i64* [[TMP26]], align 8
// CHECK6-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK6-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK6-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64* // CHECK6-NEXT: store i8* null, i8** [[TMP27]], align 8
// CHECK6-NEXT: store i64 2, i64* [[TMP28]], align 8 // CHECK6-NEXT: [[TMP28:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK6-NEXT: [[TMP29:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK6-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64*
// CHECK6-NEXT: store i64 8, i64* [[TMP29]], align 8 // CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP29]], align 8
// CHECK6-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK6-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK6-NEXT: store i8* null, i8** [[TMP30]], align 8 // CHECK6-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64*
// CHECK6-NEXT: [[TMP31:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP31]], align 8
// CHECK6-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i64* // CHECK6-NEXT: [[TMP32:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP32]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP32]], align 8
// CHECK6-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK6-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK6-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i64* // CHECK6-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i16**
// CHECK6-NEXT: store i64 [[TMP2]], i64* [[TMP34]], align 8 // CHECK6-NEXT: store i16* [[VLA]], i16** [[TMP34]], align 8
// CHECK6-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK6-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK6-NEXT: store i64 8, i64* [[TMP35]], align 8 // CHECK6-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i16**
// CHECK6-NEXT: [[TMP36:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK6-NEXT: store i16* [[VLA]], i16** [[TMP36]], align 8
// CHECK6-NEXT: store i8* null, i8** [[TMP36]], align 8 // CHECK6-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.13, i32 0, i32 4), align 8
// CHECK6-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK6-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK6-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i16** // CHECK6-NEXT: store i8* null, i8** [[TMP37]], align 8
// CHECK6-NEXT: store i16* [[VLA]], i16** [[TMP38]], align 8 // CHECK6-NEXT: [[TMP38:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK6-NEXT: [[TMP39:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK6-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64*
// CHECK6-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i16** // CHECK6-NEXT: store i64 [[TMP9]], i64* [[TMP39]], align 8
// CHECK6-NEXT: store i16* [[VLA]], i16** [[TMP40]], align 8 // CHECK6-NEXT: [[TMP40:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK6-NEXT: [[TMP41:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK6-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i64*
// CHECK6-NEXT: store i64 [[TMP12]], i64* [[TMP41]], align 8 // CHECK6-NEXT: store i64 [[TMP9]], i64* [[TMP41]], align 8
// CHECK6-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK6-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK6-NEXT: store i8* null, i8** [[TMP42]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP42]], align 8
// CHECK6-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK6-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i64* // CHECK6-NEXT: [[TMP44:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: store i64 [[TMP9]], i64* [[TMP44]], align 8
// CHECK6-NEXT: [[TMP45:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK6-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i64*
// CHECK6-NEXT: store i64 [[TMP9]], i64* [[TMP46]], align 8
// CHECK6-NEXT: [[TMP47:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5
// CHECK6-NEXT: store i64 1, i64* [[TMP47]], align 8
// CHECK6-NEXT: [[TMP48:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK6-NEXT: store i8* null, i8** [[TMP48]], align 8
// CHECK6-NEXT: [[TMP49:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP50:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP51:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK6-NEXT: [[TMP52:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 6, i8** [[TMP49]], i8** [[TMP50]], i64* [[TMP51]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK6-NEXT: [[TMP45:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 6, i8** [[TMP43]], i8** [[TMP44]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK6-NEXT: [[TMP53:%.*]] = icmp ne i32 [[TMP52]], 0 // CHECK6-NEXT: [[TMP46:%.*]] = icmp ne i32 [[TMP45]], 0
// CHECK6-NEXT: br i1 [[TMP53]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK6-NEXT: br i1 [[TMP46]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK6: omp_offload.failed: // CHECK6: omp_offload.failed:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK6: omp_offload.cont: // CHECK6: omp_offload.cont:
// CHECK6-NEXT: br label [[OMP_IF_END:%.*]] // CHECK6-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK6: omp_if.else: // CHECK6: omp_if.else:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i64 [[TMP7]], i64 2, i64 [[TMP2]], i16* [[VLA]], i64 [[TMP9]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_IF_END]] // CHECK6-NEXT: br label [[OMP_IF_END]]
// CHECK6: omp_if.end: // CHECK6: omp_if.end:
// CHECK6-NEXT: [[TMP54:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK6-NEXT: [[TMP47:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK6-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP54]] // CHECK6-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i64 [[TMP47]]
// CHECK6-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1 // CHECK6-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i64 1
// CHECK6-NEXT: [[TMP55:%.*]] = load i16, i16* [[ARRAYIDX5]], align 2 // CHECK6-NEXT: [[TMP48:%.*]] = load i16, i16* [[ARRAYIDX5]], align 2
// CHECK6-NEXT: [[CONV6:%.*]] = sext i16 [[TMP55]] to i32 // CHECK6-NEXT: [[CONV6:%.*]] = sext i16 [[TMP48]] to i32
// CHECK6-NEXT: [[TMP56:%.*]] = load i32, i32* [[B]], align 4 // CHECK6-NEXT: [[TMP49:%.*]] = load i32, i32* [[B]], align 4
// CHECK6-NEXT: [[ADD7:%.*]] = add nsw i32 [[CONV6]], [[TMP56]] // CHECK6-NEXT: [[ADD7:%.*]] = add nsw i32 [[CONV6]], [[TMP49]]
// CHECK6-NEXT: [[TMP57:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK6-NEXT: [[TMP50:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK6-NEXT: call void @llvm.stackrestore(i8* [[TMP57]]) // CHECK6-NEXT: call void @llvm.stackrestore(i8* [[TMP50]])
// CHECK6-NEXT: ret i32 [[ADD7]] // CHECK6-NEXT: ret i32 [[ADD7]]
// //
// //
// CHECK6-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK6-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK6-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] { // CHECK6-SAME: (i32 signext [[N:%.*]]) #[[ATTR0]] {
// CHECK6-NEXT: entry: // CHECK6-NEXT: entry:
// CHECK6-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1 // CHECK6-NEXT: [[ADD:%.*]] = add i32 [[SUB6]], 1
// CHECK6-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK6-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK6-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1 // CHECK6-NEXT: [[SUB7:%.*]] = sub i32 [[DIV]], 1
// CHECK6-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK6-NEXT: store i32 [[SUB7]], i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK6-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4 // CHECK6-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_5]], align 4
// CHECK6-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1 // CHECK6-NEXT: [[ADD8:%.*]] = add i32 [[TMP40]], 1
// CHECK6-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64 // CHECK6-NEXT: [[TMP41:%.*]] = zext i32 [[ADD8]] to i64
// CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK6-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK6-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK6-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK6-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK6-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK6-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK6: omp_offload.failed: // CHECK6: omp_offload.failed:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i64 [[TMP1]], i64 [[TMP3]], i64 [[TMP5]], i64 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK6: omp_offload.cont: // CHECK6: omp_offload.cont:
// CHECK6-NEXT: br label [[OMP_IF_END:%.*]] // CHECK6-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK6: omp_if.else: // CHECK6: omp_if.else:
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK6-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK6-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK6-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK6-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8 // CHECK6-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 8
// CHECK6-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK6-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK6-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK6-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK6-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK6-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK6-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK6-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK6-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK6-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK6-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK6-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK6: omp_offload.failed: // CHECK6: omp_offload.failed:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK6: omp_offload.cont: // CHECK6: omp_offload.cont:
// CHECK6-NEXT: br label [[OMP_IF_END:%.*]] // CHECK6-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK6: omp_if.else: // CHECK6: omp_if.else:
Show All 32 Lines
// CHECK6-NEXT: store i32 [[TMP4]], i32* [[CONV4]], align 4 // CHECK6-NEXT: store i32 [[TMP4]], i32* [[CONV4]], align 4
// CHECK6-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8 // CHECK6-NEXT: [[TMP5:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK6-NEXT: [[TMP6:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK6-NEXT: [[TMP6:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK6-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP6]] to i1 // CHECK6-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP6]] to i1
// CHECK6-NEXT: [[CONV5:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8* // CHECK6-NEXT: [[CONV5:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i8*
// CHECK6-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 // CHECK6-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
// CHECK6-NEXT: store i8 [[FROMBOOL]], i8* [[CONV5]], align 1 // CHECK6-NEXT: store i8 [[FROMBOOL]], i8* [[CONV5]], align 1
// CHECK6-NEXT: [[TMP7:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK6-NEXT: [[TMP7:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]], i64 [[TMP7]]) // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i64, i64, i64, i16*, i64)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i64 [[TMP5]], i64 [[TMP1]], i64 [[TMP2]], i16* [[TMP3]], i64 [[TMP7]])
// CHECK6-NEXT: ret void // CHECK6-NEXT: ret void
// //
// //
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry: // CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8 // CHECK6-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 8
// CHECK6-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 162 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2 // CHECK6-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV2]], align 2
// CHECK6-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK6-NEXT: [[CONV6:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK6-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2 // CHECK6-NEXT: store i16 [[TMP5]], i16* [[CONV6]], align 2
// CHECK6-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK6-NEXT: [[TMP6:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK6-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1 // CHECK6-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV3]], align 1
// CHECK6-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8* // CHECK6-NEXT: [[CONV7:%.*]] = bitcast i64* [[AAA_CASTED]] to i8*
// CHECK6-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1 // CHECK6-NEXT: store i8 [[TMP7]], i8* [[CONV7]], align 1
// CHECK6-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8 // CHECK6-NEXT: [[TMP8:%.*]] = load i64, i64* [[AAA_CASTED]], align 8
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK6-NEXT: ret void // CHECK6-NEXT: ret void
// //
// //
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[N:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[N:%.*]], i64 [[AA:%.*]], i64 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry: // CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[AAA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 145 Lines • ▼ Show 20 Lines
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK6-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32* // CHECK6-NEXT: [[CONV2:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK6-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4 // CHECK6-NEXT: store i32 [[TMP1]], i32* [[CONV2]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8 // CHECK6-NEXT: [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK6-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2 // CHECK6-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV1]], align 2
// CHECK6-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16* // CHECK6-NEXT: [[CONV3:%.*]] = bitcast i64* [[AA_CASTED]] to i16*
// CHECK6-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2 // CHECK6-NEXT: store i16 [[TMP3]], i16* [[CONV3]], align 2
// CHECK6-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8 // CHECK6-NEXT: [[TMP4:%.*]] = load i64, i64* [[AA_CASTED]], align 8
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i64 [[TMP2]], i64 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK6-NEXT: ret void // CHECK6-NEXT: ret void
// //
// //
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]], i64 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry: // CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK6-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK6-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS12:%.*]] = alloca [2 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS12:%.*]] = alloca [2 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_PTRS13:%.*]] = alloca [2 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS13:%.*]] = alloca [2 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4
// CHECK7-NEXT: [[_TMP15:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[_TMP15:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[A_CASTED18:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[A_CASTED18:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [9 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [9 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [9 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [9 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [9 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [9 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 4
// CHECK7-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK7-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK7-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK7-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK7-NEXT: store i32 0, i32* [[A]], align 4 // CHECK7-NEXT: store i32 0, i32* [[A]], align 4
// CHECK7-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK7-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK7-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK7-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK7-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK7-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
▲ Show 20 Lines • Show All 157 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[TMP93:%.*]] = mul nuw i32 [[TMP92]], 8 // CHECK7-NEXT: [[TMP93:%.*]] = mul nuw i32 [[TMP92]], 8
// CHECK7-NEXT: [[TMP94:%.*]] = sext i32 [[TMP93]] to i64 // CHECK7-NEXT: [[TMP94:%.*]] = sext i32 [[TMP93]] to i64
// CHECK7-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 // CHECK7-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK7-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i32* // CHECK7-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i32*
// CHECK7-NEXT: store i32 [[TMP88]], i32* [[TMP96]], align 4 // CHECK7-NEXT: store i32 [[TMP88]], i32* [[TMP96]], align 4
// CHECK7-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 // CHECK7-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK7-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32* // CHECK7-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32*
// CHECK7-NEXT: store i32 [[TMP88]], i32* [[TMP98]], align 4 // CHECK7-NEXT: store i32 [[TMP88]], i32* [[TMP98]], align 4
// CHECK7-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK7-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 0
// CHECK7-NEXT: store i64 4, i64* [[TMP99]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP99]], align 4
// CHECK7-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 0 // CHECK7-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK7-NEXT: store i8* null, i8** [[TMP100]], align 4 // CHECK7-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to [10 x float]**
// CHECK7-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1 // CHECK7-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP101]], align 4
// CHECK7-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to [10 x float]** // CHECK7-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK7-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP102]], align 4 // CHECK7-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to [10 x float]**
// CHECK7-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1 // CHECK7-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP103]], align 4
// CHECK7-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]** // CHECK7-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 1
// CHECK7-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP104]], align 4
// CHECK7-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK7-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK7-NEXT: store i64 40, i64* [[TMP105]], align 4 // CHECK7-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to i32*
// CHECK7-NEXT: [[TMP106:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 1 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP106]], align 4
// CHECK7-NEXT: store i8* null, i8** [[TMP106]], align 4 // CHECK7-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK7-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK7-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i32* // CHECK7-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i32*
// CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP108]], align 4 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP108]], align 4
// CHECK7-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2 // CHECK7-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 2
// CHECK7-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to i32* // CHECK7-NEXT: store i8* null, i8** [[TMP109]], align 4
// CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP110]], align 4 // CHECK7-NEXT: [[TMP110:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK7-NEXT: [[TMP111:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK7-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to float**
// CHECK7-NEXT: store i64 4, i64* [[TMP111]], align 4 // CHECK7-NEXT: store float* [[VLA]], float** [[TMP111]], align 4
// CHECK7-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 2 // CHECK7-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK7-NEXT: store i8* null, i8** [[TMP112]], align 4 // CHECK7-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to float**
// CHECK7-NEXT: [[TMP113:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3 // CHECK7-NEXT: store float* [[VLA]], float** [[TMP113]], align 4
// CHECK7-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float** // CHECK7-NEXT: store i64 [[TMP91]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 3), align 4
// CHECK7-NEXT: store float* [[VLA]], float** [[TMP114]], align 4 // CHECK7-NEXT: [[TMP114:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 3
// CHECK7-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3 // CHECK7-NEXT: store i8* null, i8** [[TMP114]], align 4
// CHECK7-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float** // CHECK7-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 4
// CHECK7-NEXT: store float* [[VLA]], float** [[TMP116]], align 4 // CHECK7-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]**
// CHECK7-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK7-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 4
// CHECK7-NEXT: store i64 [[TMP91]], i64* [[TMP117]], align 4 // CHECK7-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 4
// CHECK7-NEXT: [[TMP118:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 3 // CHECK7-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to [5 x [10 x double]]**
// CHECK7-NEXT: store i8* null, i8** [[TMP118]], align 4 // CHECK7-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP118]], align 4
// CHECK7-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 4 // CHECK7-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 4
// CHECK7-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to [5 x [10 x double]]** // CHECK7-NEXT: store i8* null, i8** [[TMP119]], align 4
// CHECK7-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP120]], align 4 // CHECK7-NEXT: [[TMP120:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 5
// CHECK7-NEXT: [[TMP121:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 4 // CHECK7-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i32*
// CHECK7-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to [5 x [10 x double]]** // CHECK7-NEXT: store i32 5, i32* [[TMP121]], align 4
// CHECK7-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP122]], align 4 // CHECK7-NEXT: [[TMP122:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 5
// CHECK7-NEXT: [[TMP123:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK7-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i32*
// CHECK7-NEXT: store i64 400, i64* [[TMP123]], align 4 // CHECK7-NEXT: store i32 5, i32* [[TMP123]], align 4
// CHECK7-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 4 // CHECK7-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 5
// CHECK7-NEXT: store i8* null, i8** [[TMP124]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP124]], align 4
// CHECK7-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 5 // CHECK7-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 6
// CHECK7-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32* // CHECK7-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32*
// CHECK7-NEXT: store i32 5, i32* [[TMP126]], align 4 // CHECK7-NEXT: store i32 [[TMP3]], i32* [[TMP126]], align 4
// CHECK7-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 5 // CHECK7-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 6
// CHECK7-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i32* // CHECK7-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i32*
// CHECK7-NEXT: store i32 5, i32* [[TMP128]], align 4 // CHECK7-NEXT: store i32 [[TMP3]], i32* [[TMP128]], align 4
// CHECK7-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK7-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 6
// CHECK7-NEXT: store i64 4, i64* [[TMP129]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP129]], align 4
// CHECK7-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 5 // CHECK7-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 7
// CHECK7-NEXT: store i8* null, i8** [[TMP130]], align 4 // CHECK7-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to double**
// CHECK7-NEXT: [[TMP131:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 6 // CHECK7-NEXT: store double* [[VLA1]], double** [[TMP131]], align 4
// CHECK7-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to i32* // CHECK7-NEXT: [[TMP132:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 7
// CHECK7-NEXT: store i32 [[TMP3]], i32* [[TMP132]], align 4 // CHECK7-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to double**
// CHECK7-NEXT: [[TMP133:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 6 // CHECK7-NEXT: store double* [[VLA1]], double** [[TMP133]], align 4
// CHECK7-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to i32* // CHECK7-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 7), align 4
// CHECK7-NEXT: store i32 [[TMP3]], i32* [[TMP134]], align 4 // CHECK7-NEXT: [[TMP134:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 7
// CHECK7-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK7-NEXT: store i8* null, i8** [[TMP134]], align 4
// CHECK7-NEXT: store i64 4, i64* [[TMP135]], align 4 // CHECK7-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 8
// CHECK7-NEXT: [[TMP136:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 6 // CHECK7-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT**
// CHECK7-NEXT: store i8* null, i8** [[TMP136]], align 4 // CHECK7-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 4
// CHECK7-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 7 // CHECK7-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 8
// CHECK7-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to double** // CHECK7-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT**
// CHECK7-NEXT: store double* [[VLA1]], double** [[TMP138]], align 4 // CHECK7-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 4
// CHECK7-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 7 // CHECK7-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 8
// CHECK7-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to double** // CHECK7-NEXT: store i8* null, i8** [[TMP139]], align 4
// CHECK7-NEXT: store double* [[VLA1]], double** [[TMP140]], align 4 // CHECK7-NEXT: [[TMP140:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK7-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK7-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK7-NEXT: store i64 [[TMP94]], i64* [[TMP141]], align 4
// CHECK7-NEXT: [[TMP142:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 7
// CHECK7-NEXT: store i8* null, i8** [[TMP142]], align 4
// CHECK7-NEXT: [[TMP143:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 8
// CHECK7-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to %struct.TT**
// CHECK7-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP144]], align 4
// CHECK7-NEXT: [[TMP145:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 8
// CHECK7-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to %struct.TT**
// CHECK7-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP146]], align 4
// CHECK7-NEXT: [[TMP147:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK7-NEXT: store i64 12, i64* [[TMP147]], align 4
// CHECK7-NEXT: [[TMP148:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 8
// CHECK7-NEXT: store i8* null, i8** [[TMP148]], align 4
// CHECK7-NEXT: [[TMP149:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK7-NEXT: [[TMP150:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK7-NEXT: [[TMP151:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK7-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK7-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK7-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK7-NEXT: [[TMP142:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP140]], i8** [[TMP141]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK7-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0 // CHECK7-NEXT: [[TMP143:%.*]] = icmp ne i32 [[TMP142]], 0
// CHECK7-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED25:%.*]], label [[OMP_OFFLOAD_CONT26:%.*]] // CHECK7-NEXT: br i1 [[TMP143]], label [[OMP_OFFLOAD_FAILED25:%.*]], label [[OMP_OFFLOAD_CONT26:%.*]]
// CHECK7: omp_offload.failed25: // CHECK7: omp_offload.failed25:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT26]] // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT26]]
// CHECK7: omp_offload.cont26: // CHECK7: omp_offload.cont26:
// CHECK7-NEXT: br label [[OMP_IF_END28:%.*]] // CHECK7-NEXT: br label [[OMP_IF_END28:%.*]]
// CHECK7: omp_if.else27: // CHECK7: omp_if.else27:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_IF_END28]] // CHECK7-NEXT: br label [[OMP_IF_END28]]
// CHECK7: omp_if.end28: // CHECK7: omp_if.end28:
// CHECK7-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK7-NEXT: [[TMP144:%.*]] = load i32, i32* [[A]], align 4
// CHECK7-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK7-NEXT: [[TMP145:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK7-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK7-NEXT: call void @llvm.stackrestore(i8* [[TMP145]])
// CHECK7-NEXT: ret i32 [[TMP154]] // CHECK7-NEXT: ret i32 [[TMP144]]
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97
// CHECK7-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK7-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 674 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK7-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK7-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK7-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 4 // CHECK7-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 4
// CHECK7-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [6 x i64], align 4
// CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK7-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK7-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK7-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK7-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK7-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK7-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK7-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK7-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 24 Lines
// CHECK7-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 2 // CHECK7-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 2
// CHECK7-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64 // CHECK7-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64
// CHECK7-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1** // CHECK7-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1**
// CHECK7-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 4 // CHECK7-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 4
// CHECK7-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double** // CHECK7-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double**
// CHECK7-NEXT: store double* [[A]], double** [[TMP16]], align 4 // CHECK7-NEXT: store double* [[A]], double** [[TMP16]], align 4
// CHECK7-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK7-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK7-NEXT: store i64 8, i64* [[TMP17]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK7-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK7-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK7-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK7-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK7-NEXT: store i32 [[TMP6]], i32* [[TMP19]], align 4
// CHECK7-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK7-NEXT: [[TMP20:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK7-NEXT: store i32 [[TMP6]], i32* [[TMP20]], align 4 // CHECK7-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK7-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK7-NEXT: store i32 [[TMP6]], i32* [[TMP21]], align 4
// CHECK7-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32* // CHECK7-NEXT: [[TMP22:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK7-NEXT: store i32 [[TMP6]], i32* [[TMP22]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK7-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK7-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK7-NEXT: store i64 4, i64* [[TMP23]], align 4 // CHECK7-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK7-NEXT: [[TMP24:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK7-NEXT: store i32 2, i32* [[TMP24]], align 4
// CHECK7-NEXT: store i8* null, i8** [[TMP24]], align 4 // CHECK7-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK7-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK7-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK7-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK7-NEXT: store i32 2, i32* [[TMP26]], align 4 // CHECK7-NEXT: store i32 2, i32* [[TMP26]], align 4
// CHECK7-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK7-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK7-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32* // CHECK7-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK7-NEXT: store i32 2, i32* [[TMP28]], align 4 // CHECK7-NEXT: [[TMP28:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK7-NEXT: [[TMP29:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK7-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK7-NEXT: store i64 4, i64* [[TMP29]], align 4 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK7-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK7-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK7-NEXT: store i8* null, i8** [[TMP30]], align 4 // CHECK7-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32*
// CHECK7-NEXT: [[TMP31:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4
// CHECK7-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK7-NEXT: [[TMP32:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP32]], align 4
// CHECK7-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK7-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK7-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32* // CHECK7-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i16**
// CHECK7-NEXT: store i32 [[TMP1]], i32* [[TMP34]], align 4 // CHECK7-NEXT: store i16* [[VLA]], i16** [[TMP34]], align 4
// CHECK7-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK7-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK7-NEXT: store i64 4, i64* [[TMP35]], align 4 // CHECK7-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i16**
// CHECK7-NEXT: [[TMP36:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK7-NEXT: store i16* [[VLA]], i16** [[TMP36]], align 4
// CHECK7-NEXT: store i8* null, i8** [[TMP36]], align 4 // CHECK7-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.13, i32 0, i32 4), align 4
// CHECK7-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK7-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK7-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i16** // CHECK7-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK7-NEXT: store i16* [[VLA]], i16** [[TMP38]], align 4 // CHECK7-NEXT: [[TMP38:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK7-NEXT: [[TMP39:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK7-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK7-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i16** // CHECK7-NEXT: store i32 [[TMP8]], i32* [[TMP39]], align 4
// CHECK7-NEXT: store i16* [[VLA]], i16** [[TMP40]], align 4 // CHECK7-NEXT: [[TMP40:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK7-NEXT: [[TMP41:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK7-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK7-NEXT: store i64 [[TMP12]], i64* [[TMP41]], align 4 // CHECK7-NEXT: store i32 [[TMP8]], i32* [[TMP41]], align 4
// CHECK7-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK7-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK7-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK7-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK7-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK7-NEXT: [[TMP44:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK7-NEXT: store i32 [[TMP8]], i32* [[TMP44]], align 4
// CHECK7-NEXT: [[TMP45:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK7-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32*
// CHECK7-NEXT: store i32 [[TMP8]], i32* [[TMP46]], align 4
// CHECK7-NEXT: [[TMP47:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5
// CHECK7-NEXT: store i64 1, i64* [[TMP47]], align 4
// CHECK7-NEXT: [[TMP48:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK7-NEXT: store i8* null, i8** [[TMP48]], align 4
// CHECK7-NEXT: [[TMP49:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP50:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP51:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK7-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK7-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK7-NEXT: [[TMP52:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 6, i8** [[TMP49]], i8** [[TMP50]], i64* [[TMP51]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK7-NEXT: [[TMP45:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 6, i8** [[TMP43]], i8** [[TMP44]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK7-NEXT: [[TMP53:%.*]] = icmp ne i32 [[TMP52]], 0 // CHECK7-NEXT: [[TMP46:%.*]] = icmp ne i32 [[TMP45]], 0
// CHECK7-NEXT: br i1 [[TMP53]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK7-NEXT: br i1 [[TMP46]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK7: omp_offload.failed: // CHECK7: omp_offload.failed:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK7: omp_offload.cont: // CHECK7: omp_offload.cont:
// CHECK7-NEXT: br label [[OMP_IF_END:%.*]] // CHECK7-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK7: omp_if.else: // CHECK7: omp_if.else:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_IF_END]] // CHECK7-NEXT: br label [[OMP_IF_END]]
// CHECK7: omp_if.end: // CHECK7: omp_if.end:
// CHECK7-NEXT: [[TMP54:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK7-NEXT: [[TMP47:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP54]] // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP47]]
// CHECK7-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK7-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK7-NEXT: [[TMP55:%.*]] = load i16, i16* [[ARRAYIDX4]], align 2 // CHECK7-NEXT: [[TMP48:%.*]] = load i16, i16* [[ARRAYIDX4]], align 2
// CHECK7-NEXT: [[CONV5:%.*]] = sext i16 [[TMP55]] to i32 // CHECK7-NEXT: [[CONV5:%.*]] = sext i16 [[TMP48]] to i32
// CHECK7-NEXT: [[TMP56:%.*]] = load i32, i32* [[B]], align 4 // CHECK7-NEXT: [[TMP49:%.*]] = load i32, i32* [[B]], align 4
// CHECK7-NEXT: [[ADD6:%.*]] = add nsw i32 [[CONV5]], [[TMP56]] // CHECK7-NEXT: [[ADD6:%.*]] = add nsw i32 [[CONV5]], [[TMP49]]
// CHECK7-NEXT: [[TMP57:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK7-NEXT: [[TMP50:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK7-NEXT: call void @llvm.stackrestore(i8* [[TMP57]]) // CHECK7-NEXT: call void @llvm.stackrestore(i8* [[TMP50]])
// CHECK7-NEXT: ret i32 [[ADD6]] // CHECK7-NEXT: ret i32 [[ADD6]]
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK7-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK7-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK7-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1 // CHECK7-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1
// CHECK7-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK7-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK7-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1 // CHECK7-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1
// CHECK7-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK7-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK7-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK7-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK7-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1 // CHECK7-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1
// CHECK7-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64 // CHECK7-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64
// CHECK7-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK7-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK7-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK7-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK7-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK7-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK7-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK7-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK7: omp_offload.failed: // CHECK7: omp_offload.failed:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK7: omp_offload.cont: // CHECK7: omp_offload.cont:
// CHECK7-NEXT: br label [[OMP_IF_END:%.*]] // CHECK7-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK7: omp_if.else: // CHECK7: omp_if.else:
▲ Show 20 Lines • Show All 53 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK7-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK7-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK7-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK7-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK7-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK7-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK7-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK7-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK7-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK7-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK7-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK7-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK7-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK7-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK7-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK7-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK7-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK7-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK7-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK7-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK7: omp_offload.failed: // CHECK7: omp_offload.failed:
// CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK7-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK7-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK7: omp_offload.cont: // CHECK7: omp_offload.cont:
// CHECK7-NEXT: br label [[OMP_IF_END:%.*]] // CHECK7-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK7: omp_if.else: // CHECK7: omp_if.else:
Show All 30 Lines
// CHECK7-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK7-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK7-NEXT: [[TMP6:%.*]] = load i8, i8* [[CONV]], align 1 // CHECK7-NEXT: [[TMP6:%.*]] = load i8, i8* [[CONV]], align 1
// CHECK7-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP6]] to i1 // CHECK7-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP6]] to i1
// CHECK7-NEXT: [[CONV3:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__CASTED]] to i8* // CHECK7-NEXT: [[CONV3:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__CASTED]] to i8*
// CHECK7-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 // CHECK7-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
// CHECK7-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1 // CHECK7-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*, i32)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]], i32 [[TMP7]]) // CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*, i32)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]], i32 [[TMP7]])
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK7-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK7-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 157 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK7-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK7-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK7-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK7-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2 // CHECK7-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK7-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK7-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK7-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK7-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK7-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1 // CHECK7-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[N:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[N:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 141 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK7-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK7-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK7-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK7-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK7-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK7-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK7-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK7-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK7-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK7-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK7-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK7-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK7-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK7-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 114 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[DOTOFFLOAD_BASEPTRS12:%.*]] = alloca [2 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_BASEPTRS12:%.*]] = alloca [2 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_PTRS13:%.*]] = alloca [2 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_PTRS13:%.*]] = alloca [2 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS14:%.*]] = alloca [2 x i8*], align 4
// CHECK8-NEXT: [[_TMP15:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[_TMP15:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[A_CASTED18:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[A_CASTED18:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [9 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [9 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [9 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [9 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [9 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [9 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [9 x i64], align 4
// CHECK8-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) // CHECK8-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK8-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK8-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK8-NEXT: store i32 0, i32* [[A]], align 4 // CHECK8-NEXT: store i32 0, i32* [[A]], align 4
// CHECK8-NEXT: store i16 0, i16* [[AA]], align 2 // CHECK8-NEXT: store i16 0, i16* [[AA]], align 2
// CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK8-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK8-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK8-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4 // CHECK8-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 4
▲ Show 20 Lines • Show All 157 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[TMP93:%.*]] = mul nuw i32 [[TMP92]], 8 // CHECK8-NEXT: [[TMP93:%.*]] = mul nuw i32 [[TMP92]], 8
// CHECK8-NEXT: [[TMP94:%.*]] = sext i32 [[TMP93]] to i64 // CHECK8-NEXT: [[TMP94:%.*]] = sext i32 [[TMP93]] to i64
// CHECK8-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 // CHECK8-NEXT: [[TMP95:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK8-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i32* // CHECK8-NEXT: [[TMP96:%.*]] = bitcast i8** [[TMP95]] to i32*
// CHECK8-NEXT: store i32 [[TMP88]], i32* [[TMP96]], align 4 // CHECK8-NEXT: store i32 [[TMP88]], i32* [[TMP96]], align 4
// CHECK8-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 // CHECK8-NEXT: [[TMP97:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK8-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32* // CHECK8-NEXT: [[TMP98:%.*]] = bitcast i8** [[TMP97]] to i32*
// CHECK8-NEXT: store i32 [[TMP88]], i32* [[TMP98]], align 4 // CHECK8-NEXT: store i32 [[TMP88]], i32* [[TMP98]], align 4
// CHECK8-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK8-NEXT: [[TMP99:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 0
// CHECK8-NEXT: store i64 4, i64* [[TMP99]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP99]], align 4
// CHECK8-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 0 // CHECK8-NEXT: [[TMP100:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK8-NEXT: store i8* null, i8** [[TMP100]], align 4 // CHECK8-NEXT: [[TMP101:%.*]] = bitcast i8** [[TMP100]] to [10 x float]**
// CHECK8-NEXT: [[TMP101:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1 // CHECK8-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP101]], align 4
// CHECK8-NEXT: [[TMP102:%.*]] = bitcast i8** [[TMP101]] to [10 x float]** // CHECK8-NEXT: [[TMP102:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK8-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP102]], align 4 // CHECK8-NEXT: [[TMP103:%.*]] = bitcast i8** [[TMP102]] to [10 x float]**
// CHECK8-NEXT: [[TMP103:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1 // CHECK8-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP103]], align 4
// CHECK8-NEXT: [[TMP104:%.*]] = bitcast i8** [[TMP103]] to [10 x float]** // CHECK8-NEXT: [[TMP104:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 1
// CHECK8-NEXT: store [10 x float]* [[B]], [10 x float]** [[TMP104]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP104]], align 4
// CHECK8-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK8-NEXT: [[TMP105:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK8-NEXT: store i64 40, i64* [[TMP105]], align 4 // CHECK8-NEXT: [[TMP106:%.*]] = bitcast i8** [[TMP105]] to i32*
// CHECK8-NEXT: [[TMP106:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 1 // CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP106]], align 4
// CHECK8-NEXT: store i8* null, i8** [[TMP106]], align 4 // CHECK8-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK8-NEXT: [[TMP107:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK8-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i32* // CHECK8-NEXT: [[TMP108:%.*]] = bitcast i8** [[TMP107]] to i32*
// CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP108]], align 4 // CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP108]], align 4
// CHECK8-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2 // CHECK8-NEXT: [[TMP109:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 2
// CHECK8-NEXT: [[TMP110:%.*]] = bitcast i8** [[TMP109]] to i32* // CHECK8-NEXT: store i8* null, i8** [[TMP109]], align 4
// CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP110]], align 4 // CHECK8-NEXT: [[TMP110:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK8-NEXT: [[TMP111:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK8-NEXT: [[TMP111:%.*]] = bitcast i8** [[TMP110]] to float**
// CHECK8-NEXT: store i64 4, i64* [[TMP111]], align 4 // CHECK8-NEXT: store float* [[VLA]], float** [[TMP111]], align 4
// CHECK8-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 2 // CHECK8-NEXT: [[TMP112:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK8-NEXT: store i8* null, i8** [[TMP112]], align 4 // CHECK8-NEXT: [[TMP113:%.*]] = bitcast i8** [[TMP112]] to float**
// CHECK8-NEXT: [[TMP113:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3 // CHECK8-NEXT: store float* [[VLA]], float** [[TMP113]], align 4
// CHECK8-NEXT: [[TMP114:%.*]] = bitcast i8** [[TMP113]] to float** // CHECK8-NEXT: store i64 [[TMP91]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 3), align 4
// CHECK8-NEXT: store float* [[VLA]], float** [[TMP114]], align 4 // CHECK8-NEXT: [[TMP114:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 3
// CHECK8-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3 // CHECK8-NEXT: store i8* null, i8** [[TMP114]], align 4
// CHECK8-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to float** // CHECK8-NEXT: [[TMP115:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 4
// CHECK8-NEXT: store float* [[VLA]], float** [[TMP116]], align 4 // CHECK8-NEXT: [[TMP116:%.*]] = bitcast i8** [[TMP115]] to [5 x [10 x double]]**
// CHECK8-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK8-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP116]], align 4
// CHECK8-NEXT: store i64 [[TMP91]], i64* [[TMP117]], align 4 // CHECK8-NEXT: [[TMP117:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 4
// CHECK8-NEXT: [[TMP118:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 3 // CHECK8-NEXT: [[TMP118:%.*]] = bitcast i8** [[TMP117]] to [5 x [10 x double]]**
// CHECK8-NEXT: store i8* null, i8** [[TMP118]], align 4 // CHECK8-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP118]], align 4
// CHECK8-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 4 // CHECK8-NEXT: [[TMP119:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 4
// CHECK8-NEXT: [[TMP120:%.*]] = bitcast i8** [[TMP119]] to [5 x [10 x double]]** // CHECK8-NEXT: store i8* null, i8** [[TMP119]], align 4
// CHECK8-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP120]], align 4 // CHECK8-NEXT: [[TMP120:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 5
// CHECK8-NEXT: [[TMP121:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 4 // CHECK8-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i32*
// CHECK8-NEXT: [[TMP122:%.*]] = bitcast i8** [[TMP121]] to [5 x [10 x double]]** // CHECK8-NEXT: store i32 5, i32* [[TMP121]], align 4
// CHECK8-NEXT: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[TMP122]], align 4 // CHECK8-NEXT: [[TMP122:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 5
// CHECK8-NEXT: [[TMP123:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK8-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i32*
// CHECK8-NEXT: store i64 400, i64* [[TMP123]], align 4 // CHECK8-NEXT: store i32 5, i32* [[TMP123]], align 4
// CHECK8-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 4 // CHECK8-NEXT: [[TMP124:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 5
// CHECK8-NEXT: store i8* null, i8** [[TMP124]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP124]], align 4
// CHECK8-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 5 // CHECK8-NEXT: [[TMP125:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 6
// CHECK8-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32* // CHECK8-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32*
// CHECK8-NEXT: store i32 5, i32* [[TMP126]], align 4 // CHECK8-NEXT: store i32 [[TMP3]], i32* [[TMP126]], align 4
// CHECK8-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 5 // CHECK8-NEXT: [[TMP127:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 6
// CHECK8-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i32* // CHECK8-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i32*
// CHECK8-NEXT: store i32 5, i32* [[TMP128]], align 4 // CHECK8-NEXT: store i32 [[TMP3]], i32* [[TMP128]], align 4
// CHECK8-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK8-NEXT: [[TMP129:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 6
// CHECK8-NEXT: store i64 4, i64* [[TMP129]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP129]], align 4
// CHECK8-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 5 // CHECK8-NEXT: [[TMP130:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 7
// CHECK8-NEXT: store i8* null, i8** [[TMP130]], align 4 // CHECK8-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to double**
// CHECK8-NEXT: [[TMP131:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 6 // CHECK8-NEXT: store double* [[VLA1]], double** [[TMP131]], align 4
// CHECK8-NEXT: [[TMP132:%.*]] = bitcast i8** [[TMP131]] to i32* // CHECK8-NEXT: [[TMP132:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 7
// CHECK8-NEXT: store i32 [[TMP3]], i32* [[TMP132]], align 4 // CHECK8-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to double**
// CHECK8-NEXT: [[TMP133:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 6 // CHECK8-NEXT: store double* [[VLA1]], double** [[TMP133]], align 4
// CHECK8-NEXT: [[TMP134:%.*]] = bitcast i8** [[TMP133]] to i32* // CHECK8-NEXT: store i64 [[TMP94]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 7), align 4
// CHECK8-NEXT: store i32 [[TMP3]], i32* [[TMP134]], align 4 // CHECK8-NEXT: [[TMP134:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 7
// CHECK8-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK8-NEXT: store i8* null, i8** [[TMP134]], align 4
// CHECK8-NEXT: store i64 4, i64* [[TMP135]], align 4 // CHECK8-NEXT: [[TMP135:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 8
// CHECK8-NEXT: [[TMP136:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 6 // CHECK8-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to %struct.TT**
// CHECK8-NEXT: store i8* null, i8** [[TMP136]], align 4 // CHECK8-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP136]], align 4
// CHECK8-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 7 // CHECK8-NEXT: [[TMP137:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 8
// CHECK8-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to double** // CHECK8-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to %struct.TT**
// CHECK8-NEXT: store double* [[VLA1]], double** [[TMP138]], align 4 // CHECK8-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP138]], align 4
// CHECK8-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 7 // CHECK8-NEXT: [[TMP139:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 8
// CHECK8-NEXT: [[TMP140:%.*]] = bitcast i8** [[TMP139]] to double** // CHECK8-NEXT: store i8* null, i8** [[TMP139]], align 4
// CHECK8-NEXT: store double* [[VLA1]], double** [[TMP140]], align 4 // CHECK8-NEXT: [[TMP140:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK8-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK8-NEXT: [[TMP141:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK8-NEXT: store i64 [[TMP94]], i64* [[TMP141]], align 4
// CHECK8-NEXT: [[TMP142:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 7
// CHECK8-NEXT: store i8* null, i8** [[TMP142]], align 4
// CHECK8-NEXT: [[TMP143:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 8
// CHECK8-NEXT: [[TMP144:%.*]] = bitcast i8** [[TMP143]] to %struct.TT**
// CHECK8-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP144]], align 4
// CHECK8-NEXT: [[TMP145:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 8
// CHECK8-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to %struct.TT**
// CHECK8-NEXT: store %struct.TT* [[D]], %struct.TT** [[TMP146]], align 4
// CHECK8-NEXT: [[TMP147:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8
// CHECK8-NEXT: store i64 12, i64* [[TMP147]], align 4
// CHECK8-NEXT: [[TMP148:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i32 0, i32 8
// CHECK8-NEXT: store i8* null, i8** [[TMP148]], align 4
// CHECK8-NEXT: [[TMP149:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK8-NEXT: [[TMP150:%.*]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK8-NEXT: [[TMP151:%.*]] = getelementptr inbounds [9 x i64], [9 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK8-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK8-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK8-NEXT: [[TMP152:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP149]], i8** [[TMP150]], i64* [[TMP151]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK8-NEXT: [[TMP142:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142.region_id, i32 9, i8** [[TMP140]], i8** [[TMP141]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_sizes.10, i32 0, i32 0), i64* getelementptr inbounds ([9 x i64], [9 x i64]* @.offload_maptypes.11, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK8-NEXT: [[TMP153:%.*]] = icmp ne i32 [[TMP152]], 0 // CHECK8-NEXT: [[TMP143:%.*]] = icmp ne i32 [[TMP142]], 0
// CHECK8-NEXT: br i1 [[TMP153]], label [[OMP_OFFLOAD_FAILED25:%.*]], label [[OMP_OFFLOAD_CONT26:%.*]] // CHECK8-NEXT: br i1 [[TMP143]], label [[OMP_OFFLOAD_FAILED25:%.*]], label [[OMP_OFFLOAD_CONT26:%.*]]
// CHECK8: omp_offload.failed25: // CHECK8: omp_offload.failed25:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT26]] // CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT26]]
// CHECK8: omp_offload.cont26: // CHECK8: omp_offload.cont26:
// CHECK8-NEXT: br label [[OMP_IF_END28:%.*]] // CHECK8-NEXT: br label [[OMP_IF_END28:%.*]]
// CHECK8: omp_if.else27: // CHECK8: omp_if.else27:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l142(i32 [[TMP88]], [10 x float]* [[B]], i32 [[TMP1]], float* [[VLA]], [5 x [10 x double]]* [[C]], i32 5, i32 [[TMP3]], double* [[VLA1]], %struct.TT* [[D]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_IF_END28]] // CHECK8-NEXT: br label [[OMP_IF_END28]]
// CHECK8: omp_if.end28: // CHECK8: omp_if.end28:
// CHECK8-NEXT: [[TMP154:%.*]] = load i32, i32* [[A]], align 4 // CHECK8-NEXT: [[TMP144:%.*]] = load i32, i32* [[A]], align 4
// CHECK8-NEXT: [[TMP155:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK8-NEXT: [[TMP145:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK8-NEXT: call void @llvm.stackrestore(i8* [[TMP155]]) // CHECK8-NEXT: call void @llvm.stackrestore(i8* [[TMP145]])
// CHECK8-NEXT: ret i32 [[TMP154]] // CHECK8-NEXT: ret i32 [[TMP144]]
// //
// //
// CHECK8-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97 // CHECK8-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l97
// CHECK8-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK8-SAME: (i32 [[AA:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]], i32 [[DOTCAPTURE_EXPR_1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK8-NEXT: entry: // CHECK8-NEXT: entry:
// CHECK8-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 674 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK8-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK8-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK8-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK8-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [6 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [6 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 4 // CHECK8-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [6 x i8*], align 4
// CHECK8-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [6 x i64], align 4
// CHECK8-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK8-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4 // CHECK8-NEXT: store %struct.S1* [[THIS]], %struct.S1** [[THIS_ADDR]], align 4
// CHECK8-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK8-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK8-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4 // CHECK8-NEXT: [[THIS1:%.*]] = load %struct.S1*, %struct.S1** [[THIS_ADDR]], align 4
// CHECK8-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK8-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK8-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1 // CHECK8-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
// CHECK8-NEXT: store i32 [[ADD]], i32* [[B]], align 4 // CHECK8-NEXT: store i32 [[ADD]], i32* [[B]], align 4
// CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 24 Lines
// CHECK8-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 2 // CHECK8-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 2
// CHECK8-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64 // CHECK8-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64
// CHECK8-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1** // CHECK8-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.S1**
// CHECK8-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 4 // CHECK8-NEXT: store %struct.S1* [[THIS1]], %struct.S1** [[TMP14]], align 4
// CHECK8-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double** // CHECK8-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to double**
// CHECK8-NEXT: store double* [[A]], double** [[TMP16]], align 4 // CHECK8-NEXT: store double* [[A]], double** [[TMP16]], align 4
// CHECK8-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK8-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK8-NEXT: store i64 8, i64* [[TMP17]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK8-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP18:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK8-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK8-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK8-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK8-NEXT: store i32 [[TMP6]], i32* [[TMP19]], align 4
// CHECK8-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK8-NEXT: [[TMP20:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK8-NEXT: store i32 [[TMP6]], i32* [[TMP20]], align 4 // CHECK8-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK8-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK8-NEXT: store i32 [[TMP6]], i32* [[TMP21]], align 4
// CHECK8-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32* // CHECK8-NEXT: [[TMP22:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK8-NEXT: store i32 [[TMP6]], i32* [[TMP22]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK8-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK8-NEXT: [[TMP23:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK8-NEXT: store i64 4, i64* [[TMP23]], align 4 // CHECK8-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK8-NEXT: [[TMP24:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK8-NEXT: store i32 2, i32* [[TMP24]], align 4
// CHECK8-NEXT: store i8* null, i8** [[TMP24]], align 4 // CHECK8-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK8-NEXT: [[TMP25:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK8-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK8-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK8-NEXT: store i32 2, i32* [[TMP26]], align 4 // CHECK8-NEXT: store i32 2, i32* [[TMP26]], align 4
// CHECK8-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK8-NEXT: [[TMP27:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK8-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32* // CHECK8-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK8-NEXT: store i32 2, i32* [[TMP28]], align 4 // CHECK8-NEXT: [[TMP28:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK8-NEXT: [[TMP29:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK8-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK8-NEXT: store i64 4, i64* [[TMP29]], align 4 // CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK8-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK8-NEXT: [[TMP30:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK8-NEXT: store i8* null, i8** [[TMP30]], align 4 // CHECK8-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i32*
// CHECK8-NEXT: [[TMP31:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP31]], align 4
// CHECK8-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK8-NEXT: [[TMP32:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP32]], align 4
// CHECK8-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK8-NEXT: [[TMP33:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK8-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32* // CHECK8-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i16**
// CHECK8-NEXT: store i32 [[TMP1]], i32* [[TMP34]], align 4 // CHECK8-NEXT: store i16* [[VLA]], i16** [[TMP34]], align 4
// CHECK8-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK8-NEXT: [[TMP35:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK8-NEXT: store i64 4, i64* [[TMP35]], align 4 // CHECK8-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i16**
// CHECK8-NEXT: [[TMP36:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK8-NEXT: store i16* [[VLA]], i16** [[TMP36]], align 4
// CHECK8-NEXT: store i8* null, i8** [[TMP36]], align 4 // CHECK8-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.13, i32 0, i32 4), align 4
// CHECK8-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK8-NEXT: [[TMP37:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK8-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i16** // CHECK8-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK8-NEXT: store i16* [[VLA]], i16** [[TMP38]], align 4 // CHECK8-NEXT: [[TMP38:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK8-NEXT: [[TMP39:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK8-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK8-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i16** // CHECK8-NEXT: store i32 [[TMP8]], i32* [[TMP39]], align 4
// CHECK8-NEXT: store i16* [[VLA]], i16** [[TMP40]], align 4 // CHECK8-NEXT: [[TMP40:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK8-NEXT: [[TMP41:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK8-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK8-NEXT: store i64 [[TMP12]], i64* [[TMP41]], align 4 // CHECK8-NEXT: store i32 [[TMP8]], i32* [[TMP41]], align 4
// CHECK8-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK8-NEXT: [[TMP42:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK8-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK8-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK8-NEXT: [[TMP43:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK8-NEXT: [[TMP44:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK8-NEXT: store i32 [[TMP8]], i32* [[TMP44]], align 4
// CHECK8-NEXT: [[TMP45:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK8-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32*
// CHECK8-NEXT: store i32 [[TMP8]], i32* [[TMP46]], align 4
// CHECK8-NEXT: [[TMP47:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5
// CHECK8-NEXT: store i64 1, i64* [[TMP47]], align 4
// CHECK8-NEXT: [[TMP48:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK8-NEXT: store i8* null, i8** [[TMP48]], align 4
// CHECK8-NEXT: [[TMP49:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP50:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP51:%.*]] = getelementptr inbounds [6 x i64], [6 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK8-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK8-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK8-NEXT: [[TMP52:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 6, i8** [[TMP49]], i8** [[TMP50]], i64* [[TMP51]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK8-NEXT: [[TMP45:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215.region_id, i32 6, i8** [[TMP43]], i8** [[TMP44]], i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([6 x i64], [6 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK8-NEXT: [[TMP53:%.*]] = icmp ne i32 [[TMP52]], 0 // CHECK8-NEXT: [[TMP46:%.*]] = icmp ne i32 [[TMP45]], 0
// CHECK8-NEXT: br i1 [[TMP53]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK8-NEXT: br i1 [[TMP46]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK8: omp_offload.failed: // CHECK8: omp_offload.failed:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK8: omp_offload.cont: // CHECK8: omp_offload.cont:
// CHECK8-NEXT: br label [[OMP_IF_END:%.*]] // CHECK8-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK8: omp_if.else: // CHECK8: omp_if.else:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l215(%struct.S1* [[THIS1]], i32 [[TMP6]], i32 2, i32 [[TMP1]], i16* [[VLA]], i32 [[TMP8]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_IF_END]] // CHECK8-NEXT: br label [[OMP_IF_END]]
// CHECK8: omp_if.end: // CHECK8: omp_if.end:
// CHECK8-NEXT: [[TMP54:%.*]] = mul nsw i32 1, [[TMP1]] // CHECK8-NEXT: [[TMP47:%.*]] = mul nsw i32 1, [[TMP1]]
// CHECK8-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP54]] // CHECK8-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, i16* [[VLA]], i32 [[TMP47]]
// CHECK8-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1 // CHECK8-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i16, i16* [[ARRAYIDX]], i32 1
// CHECK8-NEXT: [[TMP55:%.*]] = load i16, i16* [[ARRAYIDX4]], align 2 // CHECK8-NEXT: [[TMP48:%.*]] = load i16, i16* [[ARRAYIDX4]], align 2
// CHECK8-NEXT: [[CONV5:%.*]] = sext i16 [[TMP55]] to i32 // CHECK8-NEXT: [[CONV5:%.*]] = sext i16 [[TMP48]] to i32
// CHECK8-NEXT: [[TMP56:%.*]] = load i32, i32* [[B]], align 4 // CHECK8-NEXT: [[TMP49:%.*]] = load i32, i32* [[B]], align 4
// CHECK8-NEXT: [[ADD6:%.*]] = add nsw i32 [[CONV5]], [[TMP56]] // CHECK8-NEXT: [[ADD6:%.*]] = add nsw i32 [[CONV5]], [[TMP49]]
// CHECK8-NEXT: [[TMP57:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK8-NEXT: [[TMP50:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK8-NEXT: call void @llvm.stackrestore(i8* [[TMP57]]) // CHECK8-NEXT: call void @llvm.stackrestore(i8* [[TMP50]])
// CHECK8-NEXT: ret i32 [[ADD6]] // CHECK8-NEXT: ret i32 [[ADD6]]
// //
// //
// CHECK8-LABEL: define {{[^@]+}}@_ZL7fstatici // CHECK8-LABEL: define {{[^@]+}}@_ZL7fstatici
// CHECK8-SAME: (i32 [[N:%.*]]) #[[ATTR0]] { // CHECK8-SAME: (i32 [[N:%.*]]) #[[ATTR0]] {
// CHECK8-NEXT: entry: // CHECK8-NEXT: entry:
// CHECK8-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[A:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[A:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1 // CHECK8-NEXT: [[ADD:%.*]] = add i32 [[SUB4]], 1
// CHECK8-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1 // CHECK8-NEXT: [[DIV:%.*]] = udiv i32 [[ADD]], 1
// CHECK8-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1 // CHECK8-NEXT: [[SUB5:%.*]] = sub i32 [[DIV]], 1
// CHECK8-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK8-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK8-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK8-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK8-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1 // CHECK8-NEXT: [[ADD6:%.*]] = add i32 [[TMP40]], 1
// CHECK8-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64 // CHECK8-NEXT: [[TMP41:%.*]] = zext i32 [[ADD6]] to i64
// CHECK8-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]]) // CHECK8-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP41]])
// CHECK8-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK8-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197.region_id, i32 5, i8** [[TMP34]], i8** [[TMP35]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK8-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 // CHECK8-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK8-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK8-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK8: omp_offload.failed: // CHECK8: omp_offload.failed:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZL7fstatici_l197(i32 [[TMP1]], i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP7]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK8: omp_offload.cont: // CHECK8: omp_offload.cont:
// CHECK8-NEXT: br label [[OMP_IF_END:%.*]] // CHECK8-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK8: omp_if.else: // CHECK8: omp_if.else:
▲ Show 20 Lines • Show All 53 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK8-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK8-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]** // CHECK8-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [10 x i32]**
// CHECK8-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4 // CHECK8-NEXT: store [10 x i32]* [[B]], [10 x i32]** [[TMP18]], align 4
// CHECK8-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK8-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK8-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK8-NEXT: store i8* null, i8** [[TMP19]], align 4
// CHECK8-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK8-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK8-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK8-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK8-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK8-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK8-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180.region_id, i32 3, i8** [[TMP20]], i8** [[TMP21]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.19, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.20, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK8-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK8-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK8-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK8-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK8: omp_offload.failed: // CHECK8: omp_offload.failed:
// CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]] // CHECK8-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l180(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[B]]) #[[ATTR4]]
// CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK8-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK8: omp_offload.cont: // CHECK8: omp_offload.cont:
// CHECK8-NEXT: br label [[OMP_IF_END:%.*]] // CHECK8-NEXT: br label [[OMP_IF_END:%.*]]
// CHECK8: omp_if.else: // CHECK8: omp_if.else:
Show All 30 Lines
// CHECK8-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4 // CHECK8-NEXT: store i32 [[TMP4]], i32* [[B_CASTED]], align 4
// CHECK8-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4 // CHECK8-NEXT: [[TMP5:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK8-NEXT: [[TMP6:%.*]] = load i8, i8* [[CONV]], align 1 // CHECK8-NEXT: [[TMP6:%.*]] = load i8, i8* [[CONV]], align 1
// CHECK8-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP6]] to i1 // CHECK8-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP6]] to i1
// CHECK8-NEXT: [[CONV3:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__CASTED]] to i8* // CHECK8-NEXT: [[CONV3:%.*]] = bitcast i32* [[DOTCAPTURE_EXPR__CASTED]] to i8*
// CHECK8-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8 // CHECK8-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL]] to i8
// CHECK8-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1 // CHECK8-NEXT: store i8 [[FROMBOOL]], i8* [[CONV3]], align 1
// CHECK8-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK8-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*, i32)* @.omp_outlined..11 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]], i32 [[TMP7]]) // CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S1*, i32, i32, i32, i16*, i32)* @.omp_outlined..12 to void (i32*, i32*, ...)*), %struct.S1* [[TMP0]], i32 [[TMP5]], i32 [[TMP1]], i32 [[TMP2]], i16* [[TMP3]], i32 [[TMP7]])
// CHECK8-NEXT: ret void // CHECK8-NEXT: ret void
// //
// //
// CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..12
// CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S1* [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i16* nonnull align 2 dereferenceable(2) [[C:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK8-NEXT: entry: // CHECK8-NEXT: entry:
// CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4 // CHECK8-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S1*, align 4
// CHECK8-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 157 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK8-NEXT: [[TMP5:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK8-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK8-NEXT: [[CONV2:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK8-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2 // CHECK8-NEXT: store i16 [[TMP5]], i16* [[CONV2]], align 2
// CHECK8-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK8-NEXT: [[TMP6:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK8-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1 // CHECK8-NEXT: [[TMP7:%.*]] = load i8, i8* [[CONV1]], align 1
// CHECK8-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8* // CHECK8-NEXT: [[CONV3:%.*]] = bitcast i32* [[AAA_CASTED]] to i8*
// CHECK8-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1 // CHECK8-NEXT: store i8 [[TMP7]], i8* [[CONV3]], align 1
// CHECK8-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4 // CHECK8-NEXT: [[TMP8:%.*]] = load i32, i32* [[AAA_CASTED]], align 4
// CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]]) // CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], [10 x i32]* [[TMP0]])
// CHECK8-NEXT: ret void // CHECK8-NEXT: ret void
// //
// //
// CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..13 // CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[N:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[N:%.*]], i32 [[AA:%.*]], i32 [[AAA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK8-NEXT: entry: // CHECK8-NEXT: entry:
// CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 141 Lines • ▼ Show 20 Lines
// CHECK8-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4 // CHECK8-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[B_ADDR]], align 4
// CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4 // CHECK8-NEXT: [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK8-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4 // CHECK8-NEXT: store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK8-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4 // CHECK8-NEXT: [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK8-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2 // CHECK8-NEXT: [[TMP3:%.*]] = load i16, i16* [[CONV]], align 2
// CHECK8-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16* // CHECK8-NEXT: [[CONV1:%.*]] = bitcast i32* [[AA_CASTED]] to i16*
// CHECK8-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2 // CHECK8-NEXT: store i16 [[TMP3]], i16* [[CONV1]], align 2
// CHECK8-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4 // CHECK8-NEXT: [[TMP4:%.*]] = load i32, i32* [[AA_CASTED]], align 4
// CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]]) // CHECK8-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32 [[TMP2]], i32 [[TMP4]], [10 x i32]* [[TMP0]])
// CHECK8-NEXT: ret void // CHECK8-NEXT: ret void
// //
// //
// CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK8-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] { // CHECK8-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[A:%.*]], i32 [[AA:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR3]] {
// CHECK8-NEXT: entry: // CHECK8-NEXT: entry:
// CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK8-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK8-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK8-NEXT: [[B_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK8-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK8-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 15,784 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_simd_collapse_codegen.cpp

Show First 20 Lines • Show All 988 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK9-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK9-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK9-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK9-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK9-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK9-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK9-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK9-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK9-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK9-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK9-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR4:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR4:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK9-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP54]] // CHECK9-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 206 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR4]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR4]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK10-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK10-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK10-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK10-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK10-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK10-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK10-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK10-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK10-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK10-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK10-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK10-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK10-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK10-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK10-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK10-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR4:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR4:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK10-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP54]] // CHECK10-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 206 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR4]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR4]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK11-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK11-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK11-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK11-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK11-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK11-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK11-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK11-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK11-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR4:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR4:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK11-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP53]] // CHECK11-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 198 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR4]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR4]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 113 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK12-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK12-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK12-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK12-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK12-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK12-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK12-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK12-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK12-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK12-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK12-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK12-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK12-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK12-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK12-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK12-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR4:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR4:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK12-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP53]] // CHECK12-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 198 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR4]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67([10 x [2 x i32]]* [[A]]) #[[ATTR4]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 904 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_simd_dist_schedule_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 2,089 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS23:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS24:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES25:%.*]] = alloca [4 x i64], align 8 // CHECK9-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_28:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR4:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR4:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK9-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK9-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK9-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK9-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK9-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 8
// CHECK9-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK9-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK9-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK9-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK9-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK9-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK9-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK9-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP52]])
// CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK9-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9: omp_offload.failed15:
// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR4]]
// CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK9: omp_offload.cont16:
// CHECK9-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: [[TMP55:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: store i32 [[TMP55]], i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK9-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK9-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP60]]) // CHECK9-NEXT: store i32 [[TMP56]], i32* [[CONV19]], align 4
// CHECK9-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP57:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK9-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK9-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9: omp_offload.failed16: // CHECK9-NEXT: store i32 [[TMP58]], i32* [[CONV20]], align 4
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR4]] // CHECK9-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK9-NEXT: [[TMP60:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9: omp_offload.cont17: // CHECK9-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP63:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK9-NEXT: store i32 [[TMP63]], i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK9-NEXT: store i64 [[TMP57]], i64* [[TMP62]], align 8
// CHECK9-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32* // CHECK9-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK9-NEXT: store i32 [[TMP64]], i32* [[CONV20]], align 4 // CHECK9-NEXT: store i64 [[TMP57]], i64* [[TMP64]], align 8
// CHECK9-NEXT: [[TMP65:%.*]] = load i64, i64* [[N_CASTED19]], align 8 // CHECK9-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0
// CHECK9-NEXT: [[TMP66:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK9-NEXT: store i8* null, i8** [[TMP65]], align 8
// CHECK9-NEXT: [[CONV21:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK9-NEXT: store i32 [[TMP66]], i32* [[CONV21]], align 4 // CHECK9-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64*
// CHECK9-NEXT: [[TMP67:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP67]], align 8
// CHECK9-NEXT: [[TMP68:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK9-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK9-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK9-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i64* // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP69]], align 8
// CHECK9-NEXT: store i64 [[TMP65]], i64* [[TMP70]], align 8 // CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1
// CHECK9-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK9-NEXT: store i8* null, i8** [[TMP70]], align 8
// CHECK9-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i64* // CHECK9-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP65]], i64* [[TMP72]], align 8 // CHECK9-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32**
// CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP72]], align 8
// CHECK9-NEXT: store i64 4, i64* [[TMP73]], align 8 // CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 0 // CHECK9-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK9-NEXT: store i8* null, i8** [[TMP74]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 8
// CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP60]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i64* // CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP76]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 1 // CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK9-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i64* // CHECK9-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP78]], align 8 // CHECK9-NEXT: store i64 [[TMP59]], i64* [[TMP77]], align 8
// CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 1 // CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK9-NEXT: store i64 8, i64* [[TMP79]], align 8 // CHECK9-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP59]], i64* [[TMP79]], align 8
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3
// CHECK9-NEXT: store i8* null, i8** [[TMP80]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP80]], align 8
// CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 2 // CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32** // CHECK9-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP82]], align 8 // CHECK9-NEXT: [[TMP83:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 2 // CHECK9-NEXT: store i32 [[TMP83]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK9-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 8 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP84]], 0
// CHECK9-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 2 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: store i64 [[TMP68]], i64* [[TMP85]], align 8 // CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK9-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 2 // CHECK9-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP86]], align 8 // CHECK9-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 3 // CHECK9-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP85]], 1
// CHECK9-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK9-NEXT: [[TMP86:%.*]] = zext i32 [[ADD30]] to i64
// CHECK9-NEXT: store i64 [[TMP67]], i64* [[TMP88]], align 8 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP86]])
// CHECK9-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 3 // CHECK9-NEXT: [[TMP87:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP81]], i8** [[TMP82]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK9-NEXT: [[TMP88:%.*]] = icmp ne i32 [[TMP87]], 0
// CHECK9-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK9-NEXT: br i1 [[TMP88]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK9-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 3 // CHECK9: omp_offload.failed31:
// CHECK9-NEXT: store i64 4, i64* [[TMP91]], align 8 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i64 [[TMP57]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP59]]) #[[ATTR4]]
// CHECK9-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 3 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK9-NEXT: store i8* null, i8** [[TMP92]], align 8 // CHECK9: omp_offload.cont32:
// CHECK9-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK9-NEXT: [[TMP89:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP89]])
// CHECK9-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0
// CHECK9-NEXT: [[TMP96:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP96]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP97]], 0
// CHECK9-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK9-NEXT: [[SUB31:%.*]] = sub nsw i32 [[DIV30]], 1
// CHECK9-NEXT: store i32 [[SUB31]], i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK9-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK9-NEXT: [[ADD32:%.*]] = add nsw i32 [[TMP98]], 1
// CHECK9-NEXT: [[TMP99:%.*]] = zext i32 [[ADD32]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP99]])
// CHECK9-NEXT: [[TMP100:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP93]], i8** [[TMP94]], i64* [[TMP95]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP101:%.*]] = icmp ne i32 [[TMP100]], 0
// CHECK9-NEXT: br i1 [[TMP101]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK9: omp_offload.failed33:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i64 [[TMP65]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP67]]) #[[ATTR4]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK9: omp_offload.cont34:
// CHECK9-NEXT: [[TMP102:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP102]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP103:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP90:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP103]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP90]])
// CHECK9-NEXT: [[TMP104:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP91:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP104]] // CHECK9-NEXT: ret i32 [[TMP91]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 267 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4 // CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK9-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 140 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR4]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK9: omp_offload.failed5: // CHECK9: omp_offload.failed5:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR4]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK9: omp_offload.cont6: // CHECK9: omp_offload.cont6:
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK9-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK9-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK9: omp_offload.failed11: // CHECK9: omp_offload.failed11:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR4]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK9: omp_offload.cont12: // CHECK9: omp_offload.cont12:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 94 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK10-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK10-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS23:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS24:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES25:%.*]] = alloca [4 x i64], align 8 // CHECK10-NEXT: [[_TMP24:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP26:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_28:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR4:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR4:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK10-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK10-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK10-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK10-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK10-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK10-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 8
// CHECK10-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK10-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK10-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK10-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK10-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK10-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK10-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK10-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK10-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK10-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK10-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK10-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK10-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK10-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK10-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP52]])
// CHECK10-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK10-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10: omp_offload.failed15:
// CHECK10-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR4]]
// CHECK10-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK10-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK10: omp_offload.cont16:
// CHECK10-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: [[TMP55:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: store i32 [[TMP55]], i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK10-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK10-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK10-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP60]]) // CHECK10-NEXT: store i32 [[TMP56]], i32* [[CONV19]], align 4
// CHECK10-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP57:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK10-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4
// CHECK10-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK10-NEXT: [[CONV20:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10: omp_offload.failed16: // CHECK10-NEXT: store i32 [[TMP58]], i32* [[CONV20]], align 4
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR4]] // CHECK10-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK10-NEXT: [[TMP60:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10: omp_offload.cont17: // CHECK10-NEXT: [[TMP61:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP63:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i64*
// CHECK10-NEXT: store i32 [[TMP63]], i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK10-NEXT: store i64 [[TMP57]], i64* [[TMP62]], align 8
// CHECK10-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK10-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32* // CHECK10-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK10-NEXT: store i32 [[TMP64]], i32* [[CONV20]], align 4 // CHECK10-NEXT: store i64 [[TMP57]], i64* [[TMP64]], align 8
// CHECK10-NEXT: [[TMP65:%.*]] = load i64, i64* [[N_CASTED19]], align 8 // CHECK10-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0
// CHECK10-NEXT: [[TMP66:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_18]], align 4 // CHECK10-NEXT: store i8* null, i8** [[TMP65]], align 8
// CHECK10-NEXT: [[CONV21:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1
// CHECK10-NEXT: store i32 [[TMP66]], i32* [[CONV21]], align 4 // CHECK10-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64*
// CHECK10-NEXT: [[TMP67:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP67]], align 8
// CHECK10-NEXT: [[TMP68:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1
// CHECK10-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK10-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK10-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i64* // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP69]], align 8
// CHECK10-NEXT: store i64 [[TMP65]], i64* [[TMP70]], align 8 // CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1
// CHECK10-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK10-NEXT: store i8* null, i8** [[TMP70]], align 8
// CHECK10-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i64* // CHECK10-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2
// CHECK10-NEXT: store i64 [[TMP65]], i64* [[TMP72]], align 8 // CHECK10-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32**
// CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP72]], align 8
// CHECK10-NEXT: store i64 4, i64* [[TMP73]], align 8 // CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2
// CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 0 // CHECK10-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK10-NEXT: store i8* null, i8** [[TMP74]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 8
// CHECK10-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP60]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i64* // CHECK10-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP76]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK10-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 1 // CHECK10-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3
// CHECK10-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i64* // CHECK10-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP78]], align 8 // CHECK10-NEXT: store i64 [[TMP59]], i64* [[TMP77]], align 8
// CHECK10-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 1 // CHECK10-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3
// CHECK10-NEXT: store i64 8, i64* [[TMP79]], align 8 // CHECK10-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP59]], i64* [[TMP79]], align 8
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3
// CHECK10-NEXT: store i8* null, i8** [[TMP80]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP80]], align 8
// CHECK10-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 2 // CHECK10-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32** // CHECK10-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP82]], align 8 // CHECK10-NEXT: [[TMP83:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 2 // CHECK10-NEXT: store i32 [[TMP83]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK10-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 8 // CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP84]], 0
// CHECK10-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 2 // CHECK10-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK10-NEXT: store i64 [[TMP68]], i64* [[TMP85]], align 8 // CHECK10-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK10-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 2 // CHECK10-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP86]], align 8 // CHECK10-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 3 // CHECK10-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP85]], 1
// CHECK10-NEXT: [[TMP88:%.*]] = bitcast i8** [[TMP87]] to i64* // CHECK10-NEXT: [[TMP86:%.*]] = zext i32 [[ADD30]] to i64
// CHECK10-NEXT: store i64 [[TMP67]], i64* [[TMP88]], align 8 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP86]])
// CHECK10-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 3 // CHECK10-NEXT: [[TMP87:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP81]], i8** [[TMP82]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i64* // CHECK10-NEXT: [[TMP88:%.*]] = icmp ne i32 [[TMP87]], 0
// CHECK10-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK10-NEXT: br i1 [[TMP88]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK10-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 3 // CHECK10: omp_offload.failed31:
// CHECK10-NEXT: store i64 4, i64* [[TMP91]], align 8 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i64 [[TMP57]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP59]]) #[[ATTR4]]
// CHECK10-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS24]], i64 0, i64 3 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK10-NEXT: store i8* null, i8** [[TMP92]], align 8 // CHECK10: omp_offload.cont32:
// CHECK10-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS22]], i32 0, i32 0 // CHECK10-NEXT: [[TMP89:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS23]], i32 0, i32 0 // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP89]])
// CHECK10-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES25]], i32 0, i32 0
// CHECK10-NEXT: [[TMP96:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP96]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP97]], 0
// CHECK10-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK10-NEXT: [[SUB31:%.*]] = sub nsw i32 [[DIV30]], 1
// CHECK10-NEXT: store i32 [[SUB31]], i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK10-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_28]], align 4
// CHECK10-NEXT: [[ADD32:%.*]] = add nsw i32 [[TMP98]], 1
// CHECK10-NEXT: [[TMP99:%.*]] = zext i32 [[ADD32]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP99]])
// CHECK10-NEXT: [[TMP100:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP93]], i8** [[TMP94]], i64* [[TMP95]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP101:%.*]] = icmp ne i32 [[TMP100]], 0
// CHECK10-NEXT: br i1 [[TMP101]], label [[OMP_OFFLOAD_FAILED33:%.*]], label [[OMP_OFFLOAD_CONT34:%.*]]
// CHECK10: omp_offload.failed33:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i64 [[TMP65]], i64 [[TMP1]], i32* [[VLA]], i64 [[TMP67]]) #[[ATTR4]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT34]]
// CHECK10: omp_offload.cont34:
// CHECK10-NEXT: [[TMP102:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP102]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP103:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP90:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP103]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP90]])
// CHECK10-NEXT: [[TMP104:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP91:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP104]] // CHECK10-NEXT: ret i32 [[TMP91]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 267 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4 // CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV2]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4 // CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK10-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV3:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4 // CHECK10-NEXT: store i32 [[TMP4]], i32* [[CONV3]], align 4
// CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP5:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i64 [[TMP3]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP5]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 140 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR4]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK10: omp_offload.failed5: // CHECK10: omp_offload.failed5:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR4]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK10: omp_offload.cont6: // CHECK10: omp_offload.cont6:
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP22]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK10-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK10-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK10-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK10: omp_offload.failed11: // CHECK10: omp_offload.failed11:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR4]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK10: omp_offload.cont12: // CHECK10: omp_offload.cont12:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 94 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK11-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_16:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4 // CHECK11-NEXT: [[_TMP21:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP23:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR4:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR4:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK11-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK11-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK11-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK11-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK11-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK11-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK11-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK11-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK11-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK11-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP53]])
// CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK11-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11: omp_offload.failed14:
// CHECK11-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR4]]
// CHECK11-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK11-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK11: omp_offload.cont15:
// CHECK11-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: store i32 [[TMP56]], i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK11-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK11-NEXT: store i32 [[TMP57]], i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP61]]) // CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK11-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK11-NEXT: store i32 [[TMP59]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11: omp_offload.failed15: // CHECK11-NEXT: [[TMP61:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR4]] // CHECK11-NEXT: [[TMP62:%.*]] = sext i32 [[TMP61]] to i64
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK11-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK11: omp_offload.cont16: // CHECK11-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32*
// CHECK11-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: store i32 [[TMP58]], i32* [[TMP64]], align 4
// CHECK11-NEXT: store i32 [[TMP64]], i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK11-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK11-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[N_CASTED18]], align 4 // CHECK11-NEXT: store i32 [[TMP58]], i32* [[TMP66]], align 4
// CHECK11-NEXT: [[TMP66:%.*]] = load i32, i32* [[N_CASTED18]], align 4 // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0
// CHECK11-NEXT: [[TMP67:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP67]], align 4
// CHECK11-NEXT: store i32 [[TMP67]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1
// CHECK11-NEXT: [[TMP68:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK11-NEXT: [[TMP69:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP69]], align 4
// CHECK11-NEXT: [[TMP70:%.*]] = sext i32 [[TMP69]] to i64 // CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1
// CHECK11-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK11-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK11-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32* // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP71]], align 4
// CHECK11-NEXT: store i32 [[TMP66]], i32* [[TMP72]], align 4 // CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1
// CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK11-NEXT: store i8* null, i8** [[TMP72]], align 4
// CHECK11-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32* // CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP66]], i32* [[TMP74]], align 4 // CHECK11-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP75]], align 4 // CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2
// CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0 // CHECK11-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK11-NEXT: store i8* null, i8** [[TMP76]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 4
// CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32* // CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP78]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP77]], align 4
// CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1 // CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3
// CHECK11-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK11-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP80]], align 4 // CHECK11-NEXT: store i32 [[TMP60]], i32* [[TMP79]], align 4
// CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 1 // CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3
// CHECK11-NEXT: store i64 4, i64* [[TMP81]], align 4 // CHECK11-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32*
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP60]], i32* [[TMP81]], align 4
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP82]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP82]], align 4
// CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2 // CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK11-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK11-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 4 // CHECK11-NEXT: [[TMP85:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[TMP85]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32** // CHECK11-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP86]], align 4 // CHECK11-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP86]], 0
// CHECK11-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 2 // CHECK11-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK11-NEXT: store i64 [[TMP70]], i64* [[TMP87]], align 4 // CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[DIV25]], 1
// CHECK11-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[SUB26]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP88]], align 4 // CHECK11-NEXT: [[TMP87:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3 // CHECK11-NEXT: [[ADD27:%.*]] = add nsw i32 [[TMP87]], 1
// CHECK11-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK11-NEXT: [[TMP88:%.*]] = zext i32 [[ADD27]] to i64
// CHECK11-NEXT: store i32 [[TMP68]], i32* [[TMP90]], align 4 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP88]])
// CHECK11-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 3 // CHECK11-NEXT: [[TMP89:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP83]], i8** [[TMP84]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32* // CHECK11-NEXT: [[TMP90:%.*]] = icmp ne i32 [[TMP89]], 0
// CHECK11-NEXT: store i32 [[TMP68]], i32* [[TMP92]], align 4 // CHECK11-NEXT: br i1 [[TMP90]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
// CHECK11-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 3 // CHECK11: omp_offload.failed28:
// CHECK11-NEXT: store i64 4, i64* [[TMP93]], align 4 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i32 [[TMP58]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP60]]) #[[ATTR4]]
// CHECK11-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT29]]
// CHECK11-NEXT: store i8* null, i8** [[TMP94]], align 4 // CHECK11: omp_offload.cont29:
// CHECK11-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK11-NEXT: [[TMP91:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP91]])
// CHECK11-NEXT: [[TMP97:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
// CHECK11-NEXT: [[TMP98:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP98]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP99]], 0
// CHECK11-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK11-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK11-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK11-NEXT: [[TMP100:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK11-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP100]], 1
// CHECK11-NEXT: [[TMP101:%.*]] = zext i32 [[ADD29]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP101]])
// CHECK11-NEXT: [[TMP102:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP95]], i8** [[TMP96]], i64* [[TMP97]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP103:%.*]] = icmp ne i32 [[TMP102]], 0
// CHECK11-NEXT: br i1 [[TMP103]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK11: omp_offload.failed30:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i32 [[TMP66]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP68]]) #[[ATTR4]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK11: omp_offload.cont31:
// CHECK11-NEXT: [[TMP104:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP104]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP105:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP92:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP105]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP92]])
// CHECK11-NEXT: [[TMP106:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP93:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP106]] // CHECK11-NEXT: ret i32 [[TMP93]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 255 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 137 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR4]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK11: omp_offload.failed5: // CHECK11: omp_offload.failed5:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR4]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK11: omp_offload.cont6: // CHECK11: omp_offload.cont6:
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK11-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK11-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK11-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK11: omp_offload.failed11: // CHECK11: omp_offload.failed11:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR4]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK11: omp_offload.cont12: // CHECK11: omp_offload.cont12:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 93 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK12-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK12-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_16:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_17:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4 // CHECK12-NEXT: [[_TMP21:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP23:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR4:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR4:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK12-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK12-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK12-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK12-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK12-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK12-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK12-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK12-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK12-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK12-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK12-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK12-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK12-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK12-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK12-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK12-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP53]])
// CHECK12-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK12-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12: omp_offload.failed14:
// CHECK12-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR4]]
// CHECK12-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK12-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK12: omp_offload.cont15:
// CHECK12-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: store i32 [[TMP56]], i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK12-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK12-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK12-NEXT: store i32 [[TMP57]], i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP61]]) // CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_16]], align 4
// CHECK12-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK12-NEXT: store i32 [[TMP59]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12: omp_offload.failed15: // CHECK12-NEXT: [[TMP61:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l98(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR4]] // CHECK12-NEXT: [[TMP62:%.*]] = sext i32 [[TMP61]] to i64
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK12-NEXT: [[TMP63:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK12: omp_offload.cont16: // CHECK12-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32*
// CHECK12-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: store i32 [[TMP58]], i32* [[TMP64]], align 4
// CHECK12-NEXT: store i32 [[TMP64]], i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK12-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK12-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[N_CASTED18]], align 4 // CHECK12-NEXT: store i32 [[TMP58]], i32* [[TMP66]], align 4
// CHECK12-NEXT: [[TMP66:%.*]] = load i32, i32* [[N_CASTED18]], align 4 // CHECK12-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0
// CHECK12-NEXT: [[TMP67:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_17]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP67]], align 4
// CHECK12-NEXT: store i32 [[TMP67]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1
// CHECK12-NEXT: [[TMP68:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32*
// CHECK12-NEXT: [[TMP69:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP69]], align 4
// CHECK12-NEXT: [[TMP70:%.*]] = sext i32 [[TMP69]] to i64 // CHECK12-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1
// CHECK12-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK12-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK12-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32* // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP71]], align 4
// CHECK12-NEXT: store i32 [[TMP66]], i32* [[TMP72]], align 4 // CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1
// CHECK12-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK12-NEXT: store i8* null, i8** [[TMP72]], align 4
// CHECK12-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32* // CHECK12-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP66]], i32* [[TMP74]], align 4 // CHECK12-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP74]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP75]], align 4 // CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2
// CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0 // CHECK12-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK12-NEXT: store i8* null, i8** [[TMP76]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 4
// CHECK12-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP62]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32* // CHECK12-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP78]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP77]], align 4
// CHECK12-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1 // CHECK12-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3
// CHECK12-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32* // CHECK12-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP80]], align 4 // CHECK12-NEXT: store i32 [[TMP60]], i32* [[TMP79]], align 4
// CHECK12-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 1 // CHECK12-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3
// CHECK12-NEXT: store i64 4, i64* [[TMP81]], align 4 // CHECK12-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32*
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP60]], i32* [[TMP81]], align 4
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP82]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP82]], align 4
// CHECK12-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2 // CHECK12-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK12-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32** // CHECK12-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP84]], align 4 // CHECK12-NEXT: [[TMP85:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[TMP85]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: [[TMP86:%.*]] = bitcast i8** [[TMP85]] to i32** // CHECK12-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP86]], align 4 // CHECK12-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP86]], 0
// CHECK12-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 2 // CHECK12-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK12-NEXT: store i64 [[TMP70]], i64* [[TMP87]], align 4 // CHECK12-NEXT: [[SUB26:%.*]] = sub nsw i32 [[DIV25]], 1
// CHECK12-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[SUB26]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP88]], align 4 // CHECK12-NEXT: [[TMP87:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3 // CHECK12-NEXT: [[ADD27:%.*]] = add nsw i32 [[TMP87]], 1
// CHECK12-NEXT: [[TMP90:%.*]] = bitcast i8** [[TMP89]] to i32* // CHECK12-NEXT: [[TMP88:%.*]] = zext i32 [[ADD27]] to i64
// CHECK12-NEXT: store i32 [[TMP68]], i32* [[TMP90]], align 4 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP88]])
// CHECK12-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 3 // CHECK12-NEXT: [[TMP89:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP83]], i8** [[TMP84]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP92:%.*]] = bitcast i8** [[TMP91]] to i32* // CHECK12-NEXT: [[TMP90:%.*]] = icmp ne i32 [[TMP89]], 0
// CHECK12-NEXT: store i32 [[TMP68]], i32* [[TMP92]], align 4 // CHECK12-NEXT: br i1 [[TMP90]], label [[OMP_OFFLOAD_FAILED28:%.*]], label [[OMP_OFFLOAD_CONT29:%.*]]
// CHECK12-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 3 // CHECK12: omp_offload.failed28:
// CHECK12-NEXT: store i64 4, i64* [[TMP93]], align 4 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i32 [[TMP58]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP60]]) #[[ATTR4]]
// CHECK12-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT29]]
// CHECK12-NEXT: store i8* null, i8** [[TMP94]], align 4 // CHECK12: omp_offload.cont29:
// CHECK12-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 // CHECK12-NEXT: [[TMP91:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP91]])
// CHECK12-NEXT: [[TMP97:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
// CHECK12-NEXT: [[TMP98:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP98]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP99]], 0
// CHECK12-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK12-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK12-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK12-NEXT: [[TMP100:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK12-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP100]], 1
// CHECK12-NEXT: [[TMP101:%.*]] = zext i32 [[ADD29]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP101]])
// CHECK12-NEXT: [[TMP102:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, i32 4, i8** [[TMP95]], i8** [[TMP96]], i64* [[TMP97]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP103:%.*]] = icmp ne i32 [[TMP102]], 0
// CHECK12-NEXT: br i1 [[TMP103]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK12: omp_offload.failed30:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102(i32 [[TMP66]], i32 [[TMP0]], i32* [[VLA]], i32 [[TMP68]]) #[[ATTR4]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK12: omp_offload.cont31:
// CHECK12-NEXT: [[TMP104:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP104]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP105:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP92:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP105]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP92]])
// CHECK12-NEXT: [[TMP106:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP93:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP106]] // CHECK12-NEXT: ret i32 [[TMP93]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l94
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 255 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32 [[TMP3]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP5]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR3]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 137 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR4]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK12: omp_offload.failed5: // CHECK12: omp_offload.failed5:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR4]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK12: omp_offload.cont6: // CHECK12: omp_offload.cont6:
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK12-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK12-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK12-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK12: omp_offload.failed11: // CHECK12: omp_offload.failed11:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR4]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR4]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK12: omp_offload.cont12: // CHECK12: omp_offload.cont12:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l76
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l80
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR3]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 1,325 Lines • Show Last 20 Lines

clang/test/OpenMP/target_update_codegen.cpp

Show All 37 Lines
// CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 800] // CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 800]
// CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] [i64 2] // CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] [i64 2]
// CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i[[sz]]] [i[[sz]] 4] // CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i[[sz]]] [i[[sz]] 4]
// CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 1] // CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 1]
// CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 2] // CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 2]
// CK1: [[SIZE04:@.+]] = {{.+}}constant [2 x i64] [i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64 24] // CK1: [[SIZE04:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673] // CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673]
// CK1-LABEL: _Z3fooi // CK1-LABEL: _Z3fooi
void foo(int arg) { void foo(int arg) {
int la; int la;
float lb[arg]; float lb[arg];
// Region 00 // Region 00
▲ Show 20 Lines • Show All 85 Lines • ▼ Show 20 Linesvoid foo(int arg) {
// CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK1-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK1-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK1-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK1-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK1-DAG: store [[ST]]* @gb, [[ST]]** [[CBP0]] // CK1-DAG: store [[ST]]* @gb, [[ST]]** [[CBP0]]
// CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CP0]] // CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CP0]]
// CK1-DAG: store i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE04]], i32 0, i32 0),
// CK1-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK1-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK1-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK1-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK1-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK1-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CBP1]] // CK1-DAG: store double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1), double*** [[CBP1]]
// CK1-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK1-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK1-DAG: store double* [[VAL1:%[^,]+]], double** [[CP1]] // CK1-DAG: store double* [[VAL1:%[^,]+]], double** [[CP1]]
▲ Show 20 Lines • Show All 45 Lines • ▼ Show 20 Linesstruct ST {
T foo(T arg) { T foo(T arg) {
// Region 00 // Region 00
#pragma omp target update from(b[1:3]) if(a>123) device(arg) #pragma omp target update from(b[1:3]) if(a>123) device(arg)
{arg++;} {arg++;}
return arg; return arg;
} }
}; };
// CK2: [[SIZE00:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK2: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710674] // CK2: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710674]
// CK2-LABEL: _Z3bari // CK2-LABEL: _Z3bari
int bar(int arg){ int bar(int arg){
ST<int> A; ST<int> A;
return A.foo(arg); return A.foo(arg);
} }
// Region 00 // Region 00
// CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]] // CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
// CK2: [[IFTHEN]] // CK2: [[IFTHEN]]
// CK2-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[sz:64|32]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK2-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK2-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64 // CK2-DAG: [[DEV]] = sext i32 [[DEVi32:%[^,]+]] to i64
// CK2-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}}, // CK2-DAG: [[DEVi32]] = load i32, i32* %{{[^,]+}},
// CK2-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK2-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK2-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK2-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK2-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
// CK2-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK2-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK2-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
// CK2-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]** // CK2-DAG: [[CBP0:%.+]] = bitcast i8** [[BP0]] to [[ST]]**
// CK2-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double*** // CK2-DAG: [[CP0:%.+]] = bitcast i8** [[P0]] to double***
// CK2-DAG: store [[ST]]* [[VAR0:%[^,]+]], [[ST]]** [[CBP0]] // CK2-DAG: store [[ST]]* [[VAR0:%[^,]+]], [[ST]]** [[CBP0]]
// CK2-DAG: store double** [[SEC0:%[^,]+]], double*** [[CP0]] // CK2-DAG: store double** [[SEC0:%[^,]+]], double*** [[CP0]]
// CK2-DAG: store i[[sz]] {{%.+}}, i[[sz]]* [[S0]] // CK2-DAG: store i64 {{%.+}}, i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE00]], i32 0, i32 0),
// CK2-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1 // CK2-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1
// CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK2-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double*** // CK2-DAG: [[CBP1:%.+]] = bitcast i8** [[BP1]] to double***
// CK2-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double** // CK2-DAG: [[CP1:%.+]] = bitcast i8** [[P1]] to double**
// CK2-DAG: store double** [[CBPVAL1:%[^,]+]], double*** [[CBP1]] // CK2-DAG: store double** [[CBPVAL1:%[^,]+]], double*** [[CBP1]]
▲ Show 20 Lines • Show All 288 Lines • ▼ Show 20 Lines
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}} // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
#ifdef CK9 #ifdef CK9
struct S { struct S {
double *p; double *p;
}; };
// CK9: [[SIZE00:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 8]
// CK9: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673] // CK9: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673]
// CK9-LABEL: lvalue // CK9-LABEL: lvalue
void lvalue(struct S *s, int l, int e) { void lvalue(struct S *s, int l, int e) {
// CK9-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i{{.+}} [[GSIZE:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}, [2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK9-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}, [2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}, [2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK9-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK9-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK9-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK9-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK9-DAG: [[GSIZE]] = getelementptr inbounds {{.+}}[[SIZE:%[^,]+]]
// //
// CK9-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK9-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK9-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK9-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK9-DAG: [[SIZE0:%.+]] = getelementptr inbounds {{.+}}[[SIZE]], i{{.+}} 0, i{{.+}} 1
// CK9-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to double*** // CK9-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to double***
// CK9-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double** // CK9-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double**
// CK9-DAG: store double** [[P:%.+]], double*** [[BPC0]] // CK9-DAG: store double** [[P:%.+]], double*** [[BPC0]]
// CK9-DAG: store double* [[P_VAL:%.+]], double** [[PC0]] // CK9-DAG: store double* [[P_VAL:%.+]], double** [[PC0]]
// CK9-DAG: store i{{.+}} 8, i{{.+}}* [[SIZE0]]
// CK9-DAG: [[P]] = getelementptr inbounds [[STRUCT_S:%.+]], [[STRUCT_S]]* [[S_VAL:%.+]], i32 0, i32 0 // CK9-DAG: [[P]] = getelementptr inbounds [[STRUCT_S:%.+]], [[STRUCT_S]]* [[S_VAL:%.+]], i32 0, i32 0
// CK9-DAG: [[S_VAL]] = load [[STRUCT_S]]*, [[STRUCT_S]]** [[S_ADDR:%.+]] // CK9-DAG: [[S_VAL]] = load [[STRUCT_S]]*, [[STRUCT_S]]** [[S_ADDR:%.+]]
// CK9-DAG: [[P_VAL]] = load double*, double** [[P_1:%.+]], // CK9-DAG: [[P_VAL]] = load double*, double** [[P_1:%.+]],
// CK9-DAG: [[P_1]] = getelementptr inbounds [[STRUCT_S]], [[STRUCT_S]]* [[S_VAL_2:%.+]], i32 0, i32 0 // CK9-DAG: [[P_1]] = getelementptr inbounds [[STRUCT_S]], [[STRUCT_S]]* [[S_VAL_2:%.+]], i32 0, i32 0
// CK9-DAG: [[S_VAL_2]] = load [[STRUCT_S]]*, [[STRUCT_S]]** [[S_ADDR:%.+]] // CK9-DAG: [[S_VAL_2]] = load [[STRUCT_S]]*, [[STRUCT_S]]** [[S_ADDR:%.+]]
#pragma omp target update to(*(s->p)) #pragma omp target update to(*(s->p))
*(s->p) += e; *(s->p) += e;
#pragma omp target update from(*(s->p)) #pragma omp target update from(*(s->p))
Show All 16 Lines
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}} // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
#ifdef CK10 #ifdef CK10
struct S { struct S {
double *p; double *p;
}; };
// CK10: [[SIZE00:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 8]
// CK10: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673] // CK10: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673]
// CK10-LABEL: lvalue // CK10-LABEL: lvalue
void lvalue(struct S *s, int l, int e) { void lvalue(struct S *s, int l, int e) {
// CK10-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i{{.+}} [[GSIZE:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}, [2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK10-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}, [2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}, [2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK10-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK10-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK10-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK10-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK10-DAG: [[GSIZE]] = getelementptr inbounds {{.+}}[[SIZE:%[^,]+]]
// //
// CK10-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK10-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK10-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK10-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK10-DAG: [[SIZE0:%.+]] = getelementptr inbounds {{.+}}[[SIZE]], i{{.+}} 0, i{{.+}} 1
// CK10-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to double*** // CK10-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to double***
// CK10-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double** // CK10-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double**
// CK10-DAG: store double** [[P_VAL:%.+]], double*** [[BPC0]] // CK10-DAG: store double** [[P_VAL:%.+]], double*** [[BPC0]]
// CK10-DAG: store double* [[ADD_PTR:%.+]], double** [[PC0]] // CK10-DAG: store double* [[ADD_PTR:%.+]], double** [[PC0]]
// CK10-DAG: store i{{.+}} 8, i{{.+}}* [[SIZE0]]
// CK10-64-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[S_P:%.+]], i{{.+}} [[IDX_EXT:%.+]] // CK10-64-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[S_P:%.+]], i{{.+}} [[IDX_EXT:%.+]]
// CK10-32-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[S_P:%.+]], i{{.+}} [[L_VAL:%.+]] // CK10-32-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[S_P:%.+]], i{{.+}} [[L_VAL:%.+]]
// CK10-64-DAG: [[IDX_EXT]] = sext i32 [[L_VAL:%.+]] to i64 // CK10-64-DAG: [[IDX_EXT]] = sext i32 [[L_VAL:%.+]] to i64
// CK10-DAG: [[S_P]] = load double*, double** [[P_VAL_1:%.+]] // CK10-DAG: [[S_P]] = load double*, double** [[P_VAL_1:%.+]]
// CK10-DAG: getelementptr inbounds {{.+}}, {{.+}}* [[SS_1:%.+]], i32 0, i32 0 // CK10-DAG: getelementptr inbounds {{.+}}, {{.+}}* [[SS_1:%.+]], i32 0, i32 0
// CK10-DAG: [[SS_1]] = load {{.+}}*, {{.+}}** [[S_ADDR:%.+]] // CK10-DAG: [[SS_1]] = load {{.+}}*, {{.+}}** [[S_ADDR:%.+]]
#pragma omp target update to(*(s->p+l)) #pragma omp target update to(*(s->p+l))
*(s->p+l) += e; *(s->p+l) += e;
Show All 16 Lines
// RUN: %clang_cc1 -DCK11 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK11 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}} // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
#ifdef CK11 #ifdef CK11
struct S { struct S {
double *p; double *p;
}; };
// CK11: [[SIZE00:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 8]
// CK11: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673] // CK11: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673]
// CK11-LABEL: lvalue // CK11-LABEL: lvalue
void lvalue(struct S *s, int l, int e) { void lvalue(struct S *s, int l, int e) {
// CK11-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i{{.+}} [[GSIZE:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}, [2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK11-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}, [2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}, [2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK11-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK11-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK11-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK11-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK11-DAG: [[GSIZE]] = getelementptr inbounds {{.+}}[[SIZE:%[^,]+]]
// //
// CK11-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK11-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK11-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK11-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK11-DAG: [[SIZE0:%.+]] = getelementptr inbounds {{.+}}[[SIZE]], i{{.+}} 0, i{{.+}} 1
// CK11-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to double*** // CK11-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to double***
// CK11-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double** // CK11-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double**
// CK11-DAG: store double** [[P:%.+]], double*** [[BPC0]] // CK11-DAG: store double** [[P:%.+]], double*** [[BPC0]]
// CK11-DAG: store double* [[ARRAY_IDX:%.+]], double** [[PC0]] // CK11-DAG: store double* [[ARRAY_IDX:%.+]], double** [[PC0]]
// CK11-DAG: store i{{.+}} 8, i{{.+}} [[SIZE0]]
// CK11-DAG: [[P]] = getelementptr inbounds [[STRUCT_S:%.+]], [[STRUCT_S]]* [[SS_1:%.+]], i32 0, i32 0 // CK11-DAG: [[P]] = getelementptr inbounds [[STRUCT_S:%.+]], [[STRUCT_S]]* [[SS_1:%.+]], i32 0, i32 0
// CK11-DAG: [[ARRAY_IDX]] = getelementptr inbounds double, double* [[ADD_PTR:%.+]], i{{.+}} 3 // CK11-DAG: [[ARRAY_IDX]] = getelementptr inbounds double, double* [[ADD_PTR:%.+]], i{{.+}} 3
// CK11-64-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[S_P:%.+]], i{{.+}} [[IDX_EXT:%.+]] // CK11-64-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[S_P:%.+]], i{{.+}} [[IDX_EXT:%.+]]
// CK11-32-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[S_P:%.+]], i{{.+}} [[L_VAL:%.+]] // CK11-32-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[S_P:%.+]], i{{.+}} [[L_VAL:%.+]]
// CK11-64-DAG: [[IDX_EXT]] = sext i32 [[L_VAL:%.+]] to i64 // CK11-64-DAG: [[IDX_EXT]] = sext i32 [[L_VAL:%.+]] to i64
// CK11-DAG: [[S_P]] = load double*, double** [[P_1:%.+]], // CK11-DAG: [[S_P]] = load double*, double** [[P_1:%.+]],
// CK11-DAG: [[P_1]] = getelementptr inbounds [[STRUCT_S]], [[STRUCT_S]]* [[S_ADDR:%.+]], i32 0, i32 0 // CK11-DAG: [[P_1]] = getelementptr inbounds [[STRUCT_S]], [[STRUCT_S]]* [[S_ADDR:%.+]], i32 0, i32 0
#pragma omp target update to((s->p+l)[3]) #pragma omp target update to((s->p+l)[3])
Show All 18 Lines
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}} // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
#ifdef CK12 #ifdef CK12
struct S { struct S {
double *p; double *p;
struct S *sp; struct S *sp;
}; };
// CK12: [[SIZE00:@.+]] = {{.+}}global [3 x i64] [i64 0, i64 {{4|8}}, i64 8]
// CK12: [[MTYPE00:@.+]] = {{.+}}constant [3 x i64] [i64 0, i64 281474976710672, i64 17] // CK12: [[MTYPE00:@.+]] = {{.+}}constant [3 x i64] [i64 0, i64 281474976710672, i64 17]
// CK12-LABEL: lvalue // CK12-LABEL: lvalue
void lvalue(struct S *s, int l, int e) { void lvalue(struct S *s, int l, int e) {
// CK12-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i{{.+}} [[GSIZE:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}, [3 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK12-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 3, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[3 x i{{.+}}, [3 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[3 x i{{.+}}, [3 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK12-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK12-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK12-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK12-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK12-DAG: [[GSIZE]] = getelementptr inbounds {{.+}}[[SIZE:%[^,]+]]
// //
// CK12-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2 // CK12-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 2
// CK12-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2 // CK12-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 2
// CK12-DAG: [[SIZE2:%.+]] = getelementptr inbounds {{.+}}[[SIZE]], i{{.+}} 0, i{{.+}} 2
// CK12-DAG: [[BPC2:%.+]] = bitcast i8** [[BP2]] to double*** // CK12-DAG: [[BPC2:%.+]] = bitcast i8** [[BP2]] to double***
// CK12-DAG: [[PC2:%.+]] = bitcast i8** [[P2]] to double** // CK12-DAG: [[PC2:%.+]] = bitcast i8** [[P2]] to double**
// CK12-DAG: store double** [[P_VAL:%.+]], double*** [[BPC2]] // CK12-DAG: store double** [[P_VAL:%.+]], double*** [[BPC2]]
// CK12-DAG: store double* [[SIX:%.+]], double** [[PC2]] // CK12-DAG: store double* [[SIX:%.+]], double** [[PC2]]
// CK12-DAG: store i{{.+}} 8, i{{.+}}* [[SIZE2]]
// CK12-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK12-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK12-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK12-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK12-DAG: [[SIZE1:%.+]] = getelementptr inbounds {{.+}}[[SIZE]], i{{.+}} 0, i{{.+}} 1
// CK12-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to [[STRUCT_S:%.+]]*** // CK12-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to [[STRUCT_S:%.+]]***
// CK12-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to double*** // CK12-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to double***
// CK12-DAG: store [[STRUCT_S]]** [[SP:%.+]], [[STRUCT_S]]*** [[BPC1]] // CK12-DAG: store [[STRUCT_S]]** [[SP:%.+]], [[STRUCT_S]]*** [[BPC1]]
// CK12-DAG: store double** [[P_VAL:%.+]], double*** [[PC1]] // CK12-DAG: store double** [[P_VAL:%.+]], double*** [[PC1]]
// CK12-DAG: store i{{.+}} {{4|8}}, i{{.+}}* [[SIZE1]]
// CK12-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK12-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK12-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK12-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK12-DAG: [[SIZE0:%.+]] = getelementptr inbounds {{.+}}[[SIZE]], i{{.+}} 0, i{{.+}} 0
// CK12-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to [[STRUCT_S:%.+]]** // CK12-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to [[STRUCT_S:%.+]]**
// CK12-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to [[STRUCT_S]]*** // CK12-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to [[STRUCT_S]]***
// CK12-DAG: store [[STRUCT_S]]* [[ZERO:%.+]], [[STRUCT_S]]** [[BPC0]] // CK12-DAG: store [[STRUCT_S]]* [[ZERO:%.+]], [[STRUCT_S]]** [[BPC0]]
// CK12-DAG: store [[STRUCT_S]]** [[SP]], [[STRUCT_S]]*** [[PC0]] // CK12-DAG: store [[STRUCT_S]]** [[SP]], [[STRUCT_S]]*** [[PC0]]
// CK12-DAG: store [[STRUCT_S]]** [[S:%.+]], [[STRUCT_S]]*** [[S_VAL:%.+]] // CK12-DAG: store [[STRUCT_S]]** [[S:%.+]], [[STRUCT_S]]*** [[S_VAL:%.+]]
// CK12-DAG: store i{{.+}} {{.+}}, i{{.+}}* [[SIZE0]] // CK12-DAG: store i{{.+}} {{.+}}, i{{.+}}* getelementptr inbounds ([3 x i64], [3 x i64]* [[SIZE00]], i32 0, i32 0),
// CK12-DAG: [[SP]] = getelementptr inbounds [[STRUCT_S]], [[STRUCT_S]]* [[ONE:%.+]], i32 0, i32 1 // CK12-DAG: [[SP]] = getelementptr inbounds [[STRUCT_S]], [[STRUCT_S]]* [[ONE:%.+]], i32 0, i32 1
// CK12-DAG: [[ONE]] = load [[STRUCT_S]]*, [[STRUCT_S]]** [[S:%.+]], // CK12-DAG: [[ONE]] = load [[STRUCT_S]]*, [[STRUCT_S]]** [[S:%.+]],
// CK12-DAG: [[ZERO]] = load [[STRUCT_S]]*, [[STRUCT_S]]** [[S]], // CK12-DAG: [[ZERO]] = load [[STRUCT_S]]*, [[STRUCT_S]]** [[S]],
#pragma omp target update to(*(s->sp->p)) #pragma omp target update to(*(s->sp->p))
*(s->sp->p) = e; *(s->sp->p) = e;
#pragma omp target update from(*(s->sp->p)) #pragma omp target update from(*(s->sp->p))
} }
▲ Show 20 Lines • Show All 58 Lines • ▼ Show 20 Lines
// RUN: %clang_cc1 -DCK14 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK14 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -DCK14 -verify -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -DCK14 -verify -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -DCK14 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK14 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}} // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
#ifdef CK14 #ifdef CK14
// CK14: [[SIZE00:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 8]
// CK14: [[MTYPE00:@.+]] = private {{.*}}constant [2 x i64] [i64 0, i64 281474976710673] // CK14: [[MTYPE00:@.+]] = private {{.*}}constant [2 x i64] [i64 0, i64 281474976710673]
struct SSA { struct SSA {
double *p; double *p;
double *&pr; double *&pr;
SSA(double *&pr) : pr(pr) {} SSA(double *&pr) : pr(pr) {}
}; };
struct SSB { struct SSB {
double *d; double *d;
SSA *p; SSA *p;
SSA *&pr; SSA *&pr;
SSB(SSA *&pr) : pr(pr) {} SSB(SSA *&pr) : pr(pr) {}
// CK14-LABEL: define {{.+}}foo // CK14-LABEL: define {{.+}}foo
void foo() { void foo() {
// CK14-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GSIZE:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK14-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK14-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK14-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK14-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK14-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK14-DAG: [[GSIZE]] = getelementptr inbounds {{.+}}[[SIZE:%[^,]+]]
// CK14-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK14-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK14-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK14-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK14-DAG: [[SIZE1:%.+]] = getelementptr inbounds {{.+}}[[SIZE]], i{{.+}} 0, i{{.+}} 1
// CK14-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to double*** // CK14-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to double***
// CK14-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to double** // CK14-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to double**
// CK14-DAG: store double** [[D_VAL:%.+]], double*** [[BPC1]] // CK14-DAG: store double** [[D_VAL:%.+]], double*** [[BPC1]]
// CK14-DAG: store double* [[ADD_PTR:%.+]], double** [[PC1]] // CK14-DAG: store double* [[ADD_PTR:%.+]], double** [[PC1]]
// CK14-DAG: store i64 8, i64* [[SIZE1]]
// CK14-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[ZERO:%.+]], i{{.+}} 1 // CK14-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[ZERO:%.+]], i{{.+}} 1
// CK14-DAG: [[ZERO]] = load double*, double** [[D_VAL_2:%.+]] // CK14-DAG: [[ZERO]] = load double*, double** [[D_VAL_2:%.+]]
// CK14-DAG: [[D_VAL]] = getelementptr inbounds [[SSB:%.+]], [[SSB:%.+]]* [[THIS:%.+]], i32 0, i32 0 // CK14-DAG: [[D_VAL]] = getelementptr inbounds [[SSB:%.+]], [[SSB:%.+]]* [[THIS:%.+]], i32 0, i32 0
// CK14-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK14-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK14-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK14-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK14-DAG: [[SIZE0:%.+]] = getelementptr inbounds {{.+}}[[SIZE]], i{{.+}} 0, i{{.+}} 0
// CK14-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to [[SSB]]** // CK14-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to [[SSB]]**
// CK14-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double*** // CK14-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double***
// CK14-DAG: store [[SSB]]* [[THIS]], [[SSB]]** [[BPC0]], // CK14-DAG: store [[SSB]]* [[THIS]], [[SSB]]** [[BPC0]],
// CK14-DAG: store double** [[D_VAL]], double*** [[PC0]] // CK14-DAG: store double** [[D_VAL]], double*** [[PC0]]
// CK14-DAG: store i{{.+}} [[COMPUTE_LEN:%.+]], i{{.+}}* [[SIZE0]] // CK14-DAG: store i{{.+}} [[COMPUTE_LEN:%.+]], i{{.+}}* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE00]], i32 0, i32 0),
#pragma omp target update to(*(this->d+1)) #pragma omp target update to(*(this->d+1))
*(this->d+1) = 1; *(this->d+1) = 1;
#pragma omp target update from(*(this->d+1)) #pragma omp target update from(*(this->d+1))
} }
}; };
void lvalue_member(SSA *sap) { void lvalue_member(SSA *sap) {
Show All 16 Lines
// RUN: %clang_cc1 -DCK15 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK15 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -DCK15 -verify -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -DCK15 -verify -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -DCK15 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK15 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}} // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
#ifdef CK15 #ifdef CK15
// CK15: [[SIZE00:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 {{8|4}}]
// CK15: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673] // CK15: [[MTYPE00:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673]
struct SSA { struct SSA {
double *p; double *p;
double *&pr; double *&pr;
SSA(double *&pr) : pr(pr) {} SSA(double *&pr) : pr(pr) {}
}; };
//CK-15-LABEL: lvalue_member //CK-15-LABEL: lvalue_member
void lvalue_member(SSA *sap) { void lvalue_member(SSA *sap) {
// CK15-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i64* [[GSIZE:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null) // CK15-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE00]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}}, i8** null)
// CK15-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]] // CK15-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
// CK15-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK15-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK15-DAG: [[GSIZE]] = getelementptr inbounds {{.+}}[[SIZE:%[^,]+]]
// CK15-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1 // CK15-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
// CK15-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1 // CK15-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
// CK15-DAG: [[SIZE1:%.+]] = getelementptr inbounds {{.+}}[[SIZE]], i{{.+}} 0, i{{.+}} 1
// CK15-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to double*** // CK15-DAG: [[BPC1:%.+]] = bitcast i8** [[BP1]] to double***
// CK15-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to double** // CK15-DAG: [[PC1:%.+]] = bitcast i8** [[P1]] to double**
// CK15-DAG: store double** [[P_VAL:%.+]], double*** [[BPC1]] // CK15-DAG: store double** [[P_VAL:%.+]], double*** [[BPC1]]
// CK15-DAG: store double* [[ADD_PTR:%.+]], double** [[PC1]] // CK15-DAG: store double* [[ADD_PTR:%.+]], double** [[PC1]]
// CK15-DAG: store i64 {{4|8}}, i64* [[SIZE1]]
// CK15-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[THREE:%.+]], i{{.+}} 3 // CK15-DAG: [[ADD_PTR]] = getelementptr inbounds double, double* [[THREE:%.+]], i{{.+}} 3
// CK15-DAG: [[THREE]] = load double*, double** [[P_VAL_1:%.+]] // CK15-DAG: [[THREE]] = load double*, double** [[P_VAL_1:%.+]]
// CK15-DAG: [[P_VAL]] = getelementptr inbounds [[SSA:%.+]], [[SSA:%.+]]* [[THIS:%.+]], i32 0, i32 0 // CK15-DAG: [[P_VAL]] = getelementptr inbounds [[SSA:%.+]], [[SSA:%.+]]* [[THIS:%.+]], i32 0, i32 0
// CK15-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0 // CK15-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
// CK15-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0 // CK15-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
// CK15-DAG: [[SIZE0:%.+]] = getelementptr inbounds {{.+}}[[SIZE]], i{{.+}} 0, i{{.+}} 0
// CK15-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to [[SSA]]** // CK15-DAG: [[BPC0:%.+]] = bitcast i8** [[BP0]] to [[SSA]]**
// CK15-DAG: store [[SSA]]* [[ZERO:%.+]], [[SSA]]** [[BPC0]], // CK15-DAG: store [[SSA]]* [[ZERO:%.+]], [[SSA]]** [[BPC0]],
// CK15-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double*** // CK15-DAG: [[PC0:%.+]] = bitcast i8** [[P0]] to double***
// CK15-DAG: store double** [[P_VAL]], double*** [[PC0]], // CK15-DAG: store double** [[P_VAL]], double*** [[PC0]],
// CK15-DAG: store i{{.+}} [[COMPUTE_SIZE:%.+]], i{{.+}}* [[SIZE0]] // CK15-DAG: store i{{.+}} [[COMPUTE_SIZE:%.+]], i{{.+}}* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE00]], i32 0, i32 0),
// CK15-DAG: [[COMPUTE_SIZE]] = sdiv exact i64 [[NINE:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CK15-DAG: [[COMPUTE_SIZE]] = sdiv exact i64 [[NINE:%.+]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CK15-DAG: [[NINE]] = sub i{{.+}} [[SEVEN:%.+]], [[EIGHT:%.+]] // CK15-DAG: [[NINE]] = sub i{{.+}} [[SEVEN:%.+]], [[EIGHT:%.+]]
// CK15-DAG: [[SEVEN]] = ptrtoint i8* [[SIX:%.+]] to i64 // CK15-DAG: [[SEVEN]] = ptrtoint i8* [[SIX:%.+]] to i64
// CK15-DAG: [[EIGHT]] = ptrtoint i8* [[FIVE:%.+]] to i64 // CK15-DAG: [[EIGHT]] = ptrtoint i8* [[FIVE:%.+]] to i64
// CK15-DAG: [[SIX]] = bitcast double** {{.+}} to i8* // CK15-DAG: [[SIX]] = bitcast double** {{.+}} to i8*
// CK15-DAG: [[FIVE]] = bitcast double** {{.+}} to i8* // CK15-DAG: [[FIVE]] = bitcast double** {{.+}} to i8*
// CK15-DAG: [[ZERO]] = load [[SSA]]*, [[SSA]]** %{{.+}}, // CK15-DAG: [[ZERO]] = load [[SSA]]*, [[SSA]]** %{{.+}},
#pragma omp target update to(*(3+sap->p)) #pragma omp target update to(*(3+sap->p))
▲ Show 20 Lines • Show All 334 Lines • ▼ Show 20 Lines
// RUN: %clang_cc1 -DCK21 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s // RUN: %clang_cc1 -DCK21 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY19 %s // RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=50 -fopenmp-targets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY19 %s
// SIMD-ONLY19-NOT: {{__kmpc|__tgt}} // SIMD-ONLY19-NOT: {{__kmpc|__tgt}}
#ifdef CK21 #ifdef CK21
// CK21: [[STRUCT_ST:%.+]] = type { [10 x [10 x [10 x double*]]] } // CK21: [[STRUCT_ST:%.+]] = type { [10 x [10 x [10 x double*]]] }
// CK21: [[STRUCT_DESCRIPTOR:%.+]] = type { i64, i64, i64 } // CK21: [[STRUCT_DESCRIPTOR:%.+]] = type { i64, i64, i64 }
// CK21: [[SIZE:@.+]] = private unnamed_addr global [2 x i64] zeroinitializer
// CK21: [[MTYPE:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 299067162755073] // CK21: [[MTYPE:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 299067162755073]
struct ST { struct ST {
double *dptr[10][10][10]; double *dptr[10][10][10];
// CK21: _ZN2ST3fooEv // CK21: _ZN2ST3fooEv
void foo() { void foo() {
// CK21: [[DIMS:%.+]] = alloca [4 x [[STRUCT_DESCRIPTOR]]], // CK21: [[DIMS:%.+]] = alloca [4 x [[STRUCT_DESCRIPTOR]]],
Show All 27 Lines void foo() {
// CK21: store i64 {{4|8}}, i64* [[STRIDE_3]], // CK21: store i64 {{4|8}}, i64* [[STRIDE_3]],
// CK21: [[DIM_4:%.+]] = getelementptr inbounds [4 x [[STRUCT_DESCRIPTOR]]], [4 x [[STRUCT_DESCRIPTOR]]]* [[DIMS]], {{.+}} 0, {{.+}} 3 // CK21: [[DIM_4:%.+]] = getelementptr inbounds [4 x [[STRUCT_DESCRIPTOR]]], [4 x [[STRUCT_DESCRIPTOR]]]* [[DIMS]], {{.+}} 0, {{.+}} 3
// CK21: [[OFFSET_4:%.+]] = getelementptr inbounds [[STRUCT_DESCRIPTOR]], [[STRUCT_DESCRIPTOR]]* [[DIM_4]], {{.+}} 0, {{.+}} 0 // CK21: [[OFFSET_4:%.+]] = getelementptr inbounds [[STRUCT_DESCRIPTOR]], [[STRUCT_DESCRIPTOR]]* [[DIM_4]], {{.+}} 0, {{.+}} 0
// CK21: store i64 0, i64* [[OFFSET_4]], // CK21: store i64 0, i64* [[OFFSET_4]],
// CK21: [[COUNT_4:%.+]] = getelementptr inbounds [[STRUCT_DESCRIPTOR]], [[STRUCT_DESCRIPTOR]]* [[DIM_4]], {{.+}} 0, {{.+}} 1 // CK21: [[COUNT_4:%.+]] = getelementptr inbounds [[STRUCT_DESCRIPTOR]], [[STRUCT_DESCRIPTOR]]* [[DIM_4]], {{.+}} 0, {{.+}} 1
// CK21: store i64 1, i64* [[COUNT_4]], // CK21: store i64 1, i64* [[COUNT_4]],
// CK21: [[STRIDE_4:%.+]] = getelementptr inbounds [[STRUCT_DESCRIPTOR]], [[STRUCT_DESCRIPTOR]]* [[DIM_4]], {{.+}} 0, {{.+}} 2 // CK21: [[STRIDE_4:%.+]] = getelementptr inbounds [[STRUCT_DESCRIPTOR]], [[STRUCT_DESCRIPTOR]]* [[DIM_4]], {{.+}} 0, {{.+}} 2
// CK21: store i64 {{4|8}}, i64* [[STRIDE_4]], // CK21: store i64 {{4|8}}, i64* [[STRIDE_4]],
// CK21-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i{{.+}}* [[GEPSZ:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE]]{{.+}}) // CK21-DAG: call void @__tgt_target_data_update_mapper(%struct.ident_t* @{{.+}}, i64 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE]]{{.+}}, {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE]]{{.+}})
// CK21-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP]] // CK21-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP]]
// CK21-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]] // CK21-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
// CK21-DAG: [[PC0:%.+]] = bitcast [4 x [[STRUCT_DESCRIPTOR]]]* [[DIMS]] to i8* // CK21-DAG: [[PC0:%.+]] = bitcast [4 x [[STRUCT_DESCRIPTOR]]]* [[DIMS]] to i8*
// CK21-DAG: [[PTRS:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* %.offload_ptrs, i32 0, i32 0 // CK21-DAG: [[PTRS:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* %.offload_ptrs, i32 0, i32 0
// CK21-DAG: store i8* [[PC0]], i8** [[PTRS]], // CK21-DAG: store i8* [[PC0]], i8** [[PTRS]],
#pragma omp target update to(dptr[0:2][1:3][0:4]) #pragma omp target update to(dptr[0:2][1:3][0:4])
} }
}; };
▲ Show 20 Lines • Show All 342 Lines • Show Last 20 Lines

clang/test/OpenMP/target_update_depend_codegen.cpp

Show All 31 Lines
// CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i64] [i64 800] // CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i64] [i64 800]
// CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] [i64 2] // CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i64] [i64 2]
// CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i64] [i64 4] // CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i64] [i64 4]
// CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 1] // CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i64] [i64 1]
// CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 2] // CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i64] [i64 2]
// CK1: [[SIZE04:@.+]] = {{.+}}constant [2 x i64] [i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64 24] // CK1: [[SIZE04:@.+]] = {{.+}}global [2 x i64] [i64 0, i64 24]
// CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673] // CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i64] [i64 0, i64 281474976710673]
// CK1-LABEL: _Z3fooi // CK1-LABEL: _Z3fooi
void foo(int arg) { void foo(int arg) {
int la; int la;
float lb[arg]; float lb[arg];
// CK1: alloca [1 x %struct.kmp_depend_info], // CK1: alloca [1 x %struct.kmp_depend_info],
▲ Show 20 Lines • Show All 236 Lines • ▼ Show 20 Linesvoid foo(int arg) {
// Region 04 // Region 04
// CK1: [[BP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%.+]], i32 0, i32 0 // CK1: [[BP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%.+]], i32 0, i32 0
// CK1: [[BP0_BC:%.+]] = bitcast i8** [[BP0]] to %struct.ST** // CK1: [[BP0_BC:%.+]] = bitcast i8** [[BP0]] to %struct.ST**
// CK1: store %struct.ST* @gb, %struct.ST** [[BP0_BC]], // CK1: store %struct.ST* @gb, %struct.ST** [[BP0_BC]],
// CK1: [[P0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%.+]], i32 0, i32 0 // CK1: [[P0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%.+]], i32 0, i32 0
// CK1: [[P0_BC:%.+]] = bitcast i8** [[P0]] to double*** // CK1: [[P0_BC:%.+]] = bitcast i8** [[P0]] to double***
// CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[P0_BC]], // CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[P0_BC]],
// CK1: store i64 sdiv exact (i64 sub (i64 ptrtoint (double** getelementptr (double*, double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), i32 1) to i64), i64 ptrtoint (double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1) to i64)), i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[SIZE04]], i32 0, i32 0),
// CK1: [[BP1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 1 // CK1: [[BP1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 1
// CK1: [[BP1_BC:%.+]] = bitcast i8** [[BP1]] to double*** // CK1: [[BP1_BC:%.+]] = bitcast i8** [[BP1]] to double***
// CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[BP1_BC]], // CK1: store double** getelementptr inbounds (%struct.ST, %struct.ST* @gb, i32 0, i32 1), double*** [[BP1_BC]],
// CK1: [[P1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 1 // CK1: [[P1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 1
// CK1: [[P1_BC:%.+]] = bitcast i8** [[P1]] to double** // CK1: [[P1_BC:%.+]] = bitcast i8** [[P1]] to double**
// CK1: store double* %{{.+}}, double** [[P1_BC]], // CK1: store double* %{{.+}}, double** [[P1_BC]],
// CK1: [[GEPBP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 0 // CK1: [[GEPBP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 0
// CK1: [[GEPP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 0 // CK1: [[GEPP0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 0
▲ Show 20 Lines • Show All 131 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_distribute_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 1,615 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK9-NEXT: ret i32 [[TMP33]] // CHECK9-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 114 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 // CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK10-NEXT: ret i32 [[TMP33]] // CHECK10-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 114 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK11-NEXT: ret i32 [[TMP33]] // CHECK11-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 112 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 // CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK12-NEXT: ret i32 [[TMP33]] // CHECK12-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 625 Lines • ▼ Show 20 Lines
// CHECK25-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK25-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK25-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK25-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK25-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK25-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK25-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK25-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK25-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK25-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK25-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK25-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK25-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK25-NEXT: store i32 100, i32* [[N]], align 4 // CHECK25-NEXT: store i32 100, i32* [[N]], align 4
// CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK25-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK25-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK25-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK25-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK25-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK25-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK25-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK25-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK25-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK25-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK25-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK25-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK25-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK25-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK25-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK25-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK25-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK25-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK25-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK25-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK25-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK25-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK25-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK25-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK25-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK25-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK25-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK25-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK25-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK25-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK25-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK25-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK25-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK25-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK25-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK25-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK25-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK25-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK25-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK25-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK25-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK25-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK25-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK25-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK25-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK25-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK25-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK25-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK25-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK25-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK25-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK25-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK25-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK25-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK25-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK25-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK25-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK25-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK25-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK25-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK25-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK25-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK25-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK25-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK25-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK25: omp_offload.failed: // CHECK25: omp_offload.failed:
// CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK25: omp_offload.cont: // CHECK25: omp_offload.cont:
// CHECK25-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK25-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK25-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP33]]) // CHECK25-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP29]])
// CHECK25-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK25-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK25-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK25-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK25-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK25-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK25-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK25-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK25-NEXT: ret i32 [[TMP35]] // CHECK25-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162
// CHECK25-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK25-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK25-NEXT: entry: // CHECK25-NEXT: entry:
// CHECK25-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 150 Lines • ▼ Show 20 Lines
// CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8
// CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK25-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK25-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK25-NEXT: [[TMP22:%.*]] = load i32, i32* [[TH]], align 4 // CHECK25-NEXT: [[TMP22:%.*]] = load i32, i32* [[TH]], align 4
// CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK25-NEXT: [[TMP23:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 [[TMP22]]) // CHECK25-NEXT: [[TMP23:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 [[TMP22]])
// CHECK25-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 // CHECK25-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK25-NEXT: br i1 [[TMP24]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK25-NEXT: br i1 [[TMP24]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK25: omp_offload.failed: // CHECK25: omp_offload.failed:
// CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK25: omp_offload.cont: // CHECK25: omp_offload.cont:
// CHECK25-NEXT: ret i32 0 // CHECK25-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 100 Lines • ▼ Show 20 Lines
// CHECK26-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK26-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK26-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK26-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK26-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK26-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK26-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK26-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK26-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK26-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK26-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK26-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK26-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK26-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK26-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK26-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK26-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK26-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK26-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK26-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK26-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK26-NEXT: store i32 100, i32* [[N]], align 4 // CHECK26-NEXT: store i32 100, i32* [[N]], align 4
// CHECK26-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK26-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK26-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK26-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK26-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK26-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK26-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK26-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK26-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK26-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK26-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK26-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK26-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK26-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK26-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK26-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK26-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK26-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK26-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK26-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK26-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK26-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK26-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK26-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK26-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK26-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK26-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK26-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK26-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK26-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK26-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK26-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK26-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK26-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK26-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK26-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK26-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK26-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK26-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK26-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK26-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK26-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK26-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK26-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK26-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK26-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK26-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK26-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK26-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK26-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK26-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK26-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK26-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK26-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK26-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK26-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK26-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK26-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK26-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK26-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK26-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK26-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK26-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK26-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK26-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK26-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK26-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK26-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK26-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK26-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK26-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK26-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK26-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK26-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK26-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK26-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK26-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK26-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK26-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK26: omp_offload.failed: // CHECK26: omp_offload.failed:
// CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK26: omp_offload.cont: // CHECK26: omp_offload.cont:
// CHECK26-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK26-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK26-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP33]]) // CHECK26-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP29]])
// CHECK26-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK26-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK26-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK26-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK26-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK26-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK26-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK26-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK26-NEXT: ret i32 [[TMP35]] // CHECK26-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK26-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 // CHECK26-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162
// CHECK26-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK26-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK26-NEXT: entry: // CHECK26-NEXT: entry:
// CHECK26-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK26-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 150 Lines • ▼ Show 20 Lines
// CHECK26-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 // CHECK26-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8
// CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK26-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK26-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK26-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK26-NEXT: [[TMP22:%.*]] = load i32, i32* [[TH]], align 4 // CHECK26-NEXT: [[TMP22:%.*]] = load i32, i32* [[TH]], align 4
// CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK26-NEXT: [[TMP23:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 [[TMP22]]) // CHECK26-NEXT: [[TMP23:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 [[TMP22]])
// CHECK26-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 // CHECK26-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK26-NEXT: br i1 [[TMP24]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK26-NEXT: br i1 [[TMP24]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK26: omp_offload.failed: // CHECK26: omp_offload.failed:
// CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK26: omp_offload.cont: // CHECK26: omp_offload.cont:
// CHECK26-NEXT: ret i32 0 // CHECK26-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 100 Lines • ▼ Show 20 Lines
// CHECK27-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK27-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK27-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK27-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK27-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK27-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK27-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK27-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK27-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK27-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK27-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK27-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK27-NEXT: store i32 100, i32* [[N]], align 4 // CHECK27-NEXT: store i32 100, i32* [[N]], align 4
// CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK27-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK27-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK27-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK27-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK27-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK27-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK27-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK27-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK27-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK27-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK27-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK27-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK27-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK27-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK27-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK27-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK27-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK27-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK27-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK27-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK27-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK27-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK27-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK27-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK27-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK27-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK27-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK27-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK27-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK27-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK27-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK27-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK27-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK27-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK27-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK27-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK27-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK27-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK27-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK27-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK27-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK27-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK27-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK27-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK27-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK27-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK27-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK27-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK27-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK27-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK27-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK27-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK27-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK27-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK27-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK27-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK27-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK27-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK27-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK27-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK27: omp_offload.failed: // CHECK27: omp_offload.failed:
// CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK27: omp_offload.cont: // CHECK27: omp_offload.cont:
// CHECK27-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK27-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK27-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP33]]) // CHECK27-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP29]])
// CHECK27-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK27-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK27-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK27-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK27-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK27-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK27-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK27-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK27-NEXT: ret i32 [[TMP35]] // CHECK27-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162
// CHECK27-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK27-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK27-NEXT: entry: // CHECK27-NEXT: entry:
// CHECK27-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 146 Lines • ▼ Show 20 Lines
// CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4
// CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK27-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK27-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK27-NEXT: [[TMP22:%.*]] = load i32, i32* [[TH]], align 4 // CHECK27-NEXT: [[TMP22:%.*]] = load i32, i32* [[TH]], align 4
// CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK27-NEXT: [[TMP23:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 [[TMP22]]) // CHECK27-NEXT: [[TMP23:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 [[TMP22]])
// CHECK27-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 // CHECK27-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK27-NEXT: br i1 [[TMP24]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK27-NEXT: br i1 [[TMP24]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK27: omp_offload.failed: // CHECK27: omp_offload.failed:
// CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK27: omp_offload.cont: // CHECK27: omp_offload.cont:
// CHECK27-NEXT: ret i32 0 // CHECK27-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 97 Lines • ▼ Show 20 Lines
// CHECK28-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK28-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK28-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK28-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK28-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK28-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK28-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK28-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK28-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK28-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK28-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK28-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK28-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK28-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK28-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK28-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK28-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK28-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK28-NEXT: store i32 100, i32* [[N]], align 4 // CHECK28-NEXT: store i32 100, i32* [[N]], align 4
// CHECK28-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK28-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK28-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK28-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK28-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK28-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK28-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK28-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK28-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK28-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK28-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK28-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK28-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK28-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK28-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK28-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK28-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK28-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK28-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK28-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK28-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK28-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK28-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK28-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK28-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK28-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK28-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK28-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK28-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK28-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK28-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK28-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK28-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK28-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK28-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK28-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK28-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK28-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK28-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK28-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK28-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK28-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK28-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK28-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK28-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK28-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK28-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK28-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK28-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK28-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK28-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK28-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK28-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK28-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK28-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK28-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK28-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK28-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK28-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK28-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK28-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK28-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK28-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK28-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK28-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK28-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK28-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK28-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK28-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK28-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK28-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK28-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK28-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK28-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK28-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK28-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK28: omp_offload.failed: // CHECK28: omp_offload.failed:
// CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK28: omp_offload.cont: // CHECK28: omp_offload.cont:
// CHECK28-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK28-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK28-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP33]]) // CHECK28-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP29]])
// CHECK28-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK28-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK28-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK28-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK28-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK28-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK28-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK28-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK28-NEXT: ret i32 [[TMP35]] // CHECK28-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK28-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 // CHECK28-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162
// CHECK28-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK28-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK28-NEXT: entry: // CHECK28-NEXT: entry:
// CHECK28-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK28-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 146 Lines • ▼ Show 20 Lines
// CHECK28-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 // CHECK28-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4
// CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK28-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK28-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK28-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK28-NEXT: [[TMP22:%.*]] = load i32, i32* [[TH]], align 4 // CHECK28-NEXT: [[TMP22:%.*]] = load i32, i32* [[TH]], align 4
// CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK28-NEXT: [[TMP23:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 [[TMP22]]) // CHECK28-NEXT: [[TMP23:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 [[TMP22]])
// CHECK28-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 // CHECK28-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK28-NEXT: br i1 [[TMP24]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK28-NEXT: br i1 [[TMP24]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK28: omp_offload.failed: // CHECK28: omp_offload.failed:
// CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK28: omp_offload.cont: // CHECK28: omp_offload.cont:
// CHECK28-NEXT: ret i32 0 // CHECK28-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 91 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_distribute_collapse_codegen.cpp

Show First 20 Lines • Show All 679 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK9-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK9-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK9-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK9-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK9-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK9-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK9-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK9-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK9-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK9-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK9-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK9-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP54]] // CHECK9-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 178 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 107 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK10-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK10-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK10-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK10-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK10-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK10-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK10-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK10-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK10-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK10-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK10-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK10-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK10-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK10-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK10-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK10-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK10-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP54]] // CHECK10-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 178 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 107 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK11-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK11-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK11-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK11-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK11-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK11-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK11-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK11-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK11-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK11-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP53]] // CHECK11-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 174 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 105 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK12-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK12-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK12-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK12-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK12-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK12-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK12-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK12-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK12-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK12-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK12-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK12-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK12-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK12-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK12-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK12-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK12-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP53]] // CHECK12-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l82
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 174 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK12-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK12-SAME: () #[[ATTR5:[0-9]+]] { // CHECK12-SAME: () #[[ATTR5:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: call void @__tgt_register_requires(i64 1) // CHECK12-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
//
No newline at end of file

clang/test/OpenMP/teams_distribute_dist_schedule_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 1,520 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N_CASTED17:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[_TMP22:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES23:%.*]] = alloca [3 x i64], align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK9-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK9-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK9-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK9-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK9-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 8
// CHECK9-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK9-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK9-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK9-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK9-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK9-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK9-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK9-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP52]])
// CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK9-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9: omp_offload.failed15:
// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK9: omp_offload.cont16:
// CHECK9-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: [[TMP55:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: [[CONV18:%.*]] = bitcast i64* [[N_CASTED17]] to i32*
// CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK9-NEXT: store i32 [[TMP55]], i32* [[CONV18]], align 4
// CHECK9-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK9-NEXT: [[TMP56:%.*]] = load i64, i64* [[N_CASTED17]], align 8
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP60]]) // CHECK9-NEXT: [[TMP57:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP58:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK9-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK9-NEXT: [[TMP59:%.*]] = bitcast i8** [[TMP58]] to i64*
// CHECK9-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK9-NEXT: store i64 [[TMP56]], i64* [[TMP59]], align 8
// CHECK9: omp_offload.failed16: // CHECK9-NEXT: [[TMP60:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK9-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i64*
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK9-NEXT: store i64 [[TMP56]], i64* [[TMP61]], align 8
// CHECK9: omp_offload.cont17: // CHECK9-NEXT: [[TMP62:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 0
// CHECK9-NEXT: [[TMP63:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: store i8* null, i8** [[TMP62]], align 8
// CHECK9-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32* // CHECK9-NEXT: [[TMP63:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1
// CHECK9-NEXT: store i32 [[TMP63]], i32* [[CONV19]], align 4 // CHECK9-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK9-NEXT: [[TMP64:%.*]] = load i64, i64* [[N_CASTED18]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP64]], align 8
// CHECK9-NEXT: [[TMP65:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP65:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1
// CHECK9-NEXT: [[TMP66:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK9-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i64*
// CHECK9-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64* // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP66]], align 8
// CHECK9-NEXT: store i64 [[TMP64]], i64* [[TMP67]], align 8 // CHECK9-NEXT: [[TMP67:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 1
// CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK9-NEXT: store i8* null, i8** [[TMP67]], align 8
// CHECK9-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64* // CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP64]], i64* [[TMP69]], align 8 // CHECK9-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32**
// CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 0 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP69]], align 8
// CHECK9-NEXT: store i64 4, i64* [[TMP70]], align 8 // CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2
// CHECK9-NEXT: [[TMP71:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK9-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32**
// CHECK9-NEXT: store i8* null, i8** [[TMP71]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP71]], align 8
// CHECK9-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i64* // CHECK9-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 2
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP73]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP72]], align 8
// CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK9-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64* // CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP75]], align 8 // CHECK9-NEXT: [[TMP75:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 1 // CHECK9-NEXT: store i32 [[TMP75]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK9-NEXT: store i64 8, i64* [[TMP76]], align 8 // CHECK9-NEXT: [[TMP76:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK9-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP76]], 0
// CHECK9-NEXT: store i8* null, i8** [[TMP77]], align 8 // CHECK9-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK9-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32** // CHECK9-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP79]], align 8 // CHECK9-NEXT: [[TMP77:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK9-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP77]], 1
// CHECK9-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32** // CHECK9-NEXT: [[TMP78:%.*]] = zext i32 [[ADD28]] to i64
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP81]], align 8 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP78]])
// CHECK9-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 2 // CHECK9-NEXT: [[TMP79:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP73]], i8** [[TMP74]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: store i64 [[TMP65]], i64* [[TMP82]], align 8 // CHECK9-NEXT: [[TMP80:%.*]] = icmp ne i32 [[TMP79]], 0
// CHECK9-NEXT: [[TMP83:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK9-NEXT: br i1 [[TMP80]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK9-NEXT: store i8* null, i8** [[TMP83]], align 8 // CHECK9: omp_offload.failed29:
// CHECK9-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i64 [[TMP56]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: [[TMP85:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK9-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 0 // CHECK9: omp_offload.cont30:
// CHECK9-NEXT: [[TMP87:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP81:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP87]], i32* [[DOTCAPTURE_EXPR_25]], align 4 // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP81]])
// CHECK9-NEXT: [[TMP88:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP88]], 0
// CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK9-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[TMP89:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP89]], 1
// CHECK9-NEXT: [[TMP90:%.*]] = zext i32 [[ADD30]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP90]])
// CHECK9-NEXT: [[TMP91:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP84]], i8** [[TMP85]], i64* [[TMP86]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP92:%.*]] = icmp ne i32 [[TMP91]], 0
// CHECK9-NEXT: br i1 [[TMP92]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK9: omp_offload.failed31:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i64 [[TMP64]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK9: omp_offload.cont32:
// CHECK9-NEXT: [[TMP93:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP93]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP94:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP82:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP94]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP82]])
// CHECK9-NEXT: [[TMP95:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP83:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP95]] // CHECK9-NEXT: ret i32 [[TMP83]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 228 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 128 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK9: omp_offload.failed5: // CHECK9: omp_offload.failed5:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK9: omp_offload.cont6: // CHECK9: omp_offload.cont6:
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK9-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK9-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK9: omp_offload.failed11: // CHECK9: omp_offload.failed11:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK9: omp_offload.cont12: // CHECK9: omp_offload.cont12:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 87 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK10-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK10-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N_CASTED17:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[_TMP22:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES23:%.*]] = alloca [3 x i64], align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK10-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK10-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK10-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK10-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK10-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK10-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 8
// CHECK10-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK10-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK10-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK10-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK10-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK10-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK10-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK10-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK10-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK10-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK10-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK10-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK10-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK10-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK10-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP52]])
// CHECK10-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK10-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10: omp_offload.failed15:
// CHECK10-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK10-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK10: omp_offload.cont16:
// CHECK10-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: [[TMP55:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: [[CONV18:%.*]] = bitcast i64* [[N_CASTED17]] to i32*
// CHECK10-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK10-NEXT: store i32 [[TMP55]], i32* [[CONV18]], align 4
// CHECK10-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK10-NEXT: [[TMP56:%.*]] = load i64, i64* [[N_CASTED17]], align 8
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP60]]) // CHECK10-NEXT: [[TMP57:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP58:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK10-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK10-NEXT: [[TMP59:%.*]] = bitcast i8** [[TMP58]] to i64*
// CHECK10-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK10-NEXT: store i64 [[TMP56]], i64* [[TMP59]], align 8
// CHECK10: omp_offload.failed16: // CHECK10-NEXT: [[TMP60:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK10-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i64*
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK10-NEXT: store i64 [[TMP56]], i64* [[TMP61]], align 8
// CHECK10: omp_offload.cont17: // CHECK10-NEXT: [[TMP62:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 0
// CHECK10-NEXT: [[TMP63:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: store i8* null, i8** [[TMP62]], align 8
// CHECK10-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32* // CHECK10-NEXT: [[TMP63:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1
// CHECK10-NEXT: store i32 [[TMP63]], i32* [[CONV19]], align 4 // CHECK10-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK10-NEXT: [[TMP64:%.*]] = load i64, i64* [[N_CASTED18]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP64]], align 8
// CHECK10-NEXT: [[TMP65:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP65:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1
// CHECK10-NEXT: [[TMP66:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK10-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i64*
// CHECK10-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64* // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP66]], align 8
// CHECK10-NEXT: store i64 [[TMP64]], i64* [[TMP67]], align 8 // CHECK10-NEXT: [[TMP67:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 1
// CHECK10-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK10-NEXT: store i8* null, i8** [[TMP67]], align 8
// CHECK10-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64* // CHECK10-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2
// CHECK10-NEXT: store i64 [[TMP64]], i64* [[TMP69]], align 8 // CHECK10-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32**
// CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 0 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP69]], align 8
// CHECK10-NEXT: store i64 4, i64* [[TMP70]], align 8 // CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2
// CHECK10-NEXT: [[TMP71:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK10-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32**
// CHECK10-NEXT: store i8* null, i8** [[TMP71]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP71]], align 8
// CHECK10-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i64* // CHECK10-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 2
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP73]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP72]], align 8
// CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK10-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64* // CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP75]], align 8 // CHECK10-NEXT: [[TMP75:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 1 // CHECK10-NEXT: store i32 [[TMP75]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK10-NEXT: store i64 8, i64* [[TMP76]], align 8 // CHECK10-NEXT: [[TMP76:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK10-NEXT: [[TMP77:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK10-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP76]], 0
// CHECK10-NEXT: store i8* null, i8** [[TMP77]], align 8 // CHECK10-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK10-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2 // CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK10-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32** // CHECK10-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP79]], align 8 // CHECK10-NEXT: [[TMP77:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK10-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP77]], 1
// CHECK10-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32** // CHECK10-NEXT: [[TMP78:%.*]] = zext i32 [[ADD28]] to i64
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP81]], align 8 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP78]])
// CHECK10-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 2 // CHECK10-NEXT: [[TMP79:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP73]], i8** [[TMP74]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: store i64 [[TMP65]], i64* [[TMP82]], align 8 // CHECK10-NEXT: [[TMP80:%.*]] = icmp ne i32 [[TMP79]], 0
// CHECK10-NEXT: [[TMP83:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK10-NEXT: br i1 [[TMP80]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK10-NEXT: store i8* null, i8** [[TMP83]], align 8 // CHECK10: omp_offload.failed29:
// CHECK10-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i64 [[TMP56]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: [[TMP85:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK10-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 0 // CHECK10: omp_offload.cont30:
// CHECK10-NEXT: [[TMP87:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP81:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i32 [[TMP87]], i32* [[DOTCAPTURE_EXPR_25]], align 4 // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP81]])
// CHECK10-NEXT: [[TMP88:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP88]], 0
// CHECK10-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK10-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK10-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[TMP89:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP89]], 1
// CHECK10-NEXT: [[TMP90:%.*]] = zext i32 [[ADD30]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP90]])
// CHECK10-NEXT: [[TMP91:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP84]], i8** [[TMP85]], i64* [[TMP86]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP92:%.*]] = icmp ne i32 [[TMP91]], 0
// CHECK10-NEXT: br i1 [[TMP92]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK10: omp_offload.failed31:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i64 [[TMP64]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK10: omp_offload.cont32:
// CHECK10-NEXT: [[TMP93:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP93]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP94:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP82:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP94]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP82]])
// CHECK10-NEXT: [[TMP95:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP83:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP95]] // CHECK10-NEXT: ret i32 [[TMP83]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 228 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 128 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK10: omp_offload.failed5: // CHECK10: omp_offload.failed5:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK10: omp_offload.cont6: // CHECK10: omp_offload.cont6:
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP22]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK10-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK10-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK10-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK10: omp_offload.failed11: // CHECK10: omp_offload.failed11:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK10: omp_offload.cont12: // CHECK10: omp_offload.cont12:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 87 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK11-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED16:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS17:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS18:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS19:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[_TMP20:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES21:%.*]] = alloca [3 x i64], align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_21:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP22:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK11-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK11-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK11-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK11-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK11-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK11-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK11-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK11-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK11-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK11-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP53]])
// CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK11-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11: omp_offload.failed14:
// CHECK11-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK11-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK11: omp_offload.cont15:
// CHECK11-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: store i32 [[TMP56]], i32* [[N_CASTED16]], align 4
// CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK11-NEXT: [[TMP57:%.*]] = load i32, i32* [[N_CASTED16]], align 4
// CHECK11-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK11-NEXT: [[TMP58:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP61]]) // CHECK11-NEXT: [[TMP59:%.*]] = sext i32 [[TMP58]] to i64
// CHECK11-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP60:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK11-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK11-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i32*
// CHECK11-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK11-NEXT: store i32 [[TMP57]], i32* [[TMP61]], align 4
// CHECK11: omp_offload.failed15: // CHECK11-NEXT: [[TMP62:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK11-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32*
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK11-NEXT: store i32 [[TMP57]], i32* [[TMP63]], align 4
// CHECK11: omp_offload.cont16: // CHECK11-NEXT: [[TMP64:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 0
// CHECK11-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP64]], align 4
// CHECK11-NEXT: store i32 [[TMP64]], i32* [[N_CASTED17]], align 4 // CHECK11-NEXT: [[TMP65:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 1
// CHECK11-NEXT: [[TMP65:%.*]] = load i32, i32* [[N_CASTED17]], align 4 // CHECK11-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK11-NEXT: [[TMP66:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP66]], align 4
// CHECK11-NEXT: [[TMP67:%.*]] = sext i32 [[TMP66]] to i64 // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 1
// CHECK11-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK11-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i32*
// CHECK11-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32* // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP68]], align 4
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[TMP69]], align 4 // CHECK11-NEXT: [[TMP69:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 1
// CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK11-NEXT: store i8* null, i8** [[TMP69]], align 4
// CHECK11-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32* // CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[TMP71]], align 4 // CHECK11-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32**
// CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP71]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP72]], align 4 // CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 2
// CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0 // CHECK11-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32**
// CHECK11-NEXT: store i8* null, i8** [[TMP73]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP73]], align 4
// CHECK11-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP59]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32* // CHECK11-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP75]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP74]], align 4
// CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1 // CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK11-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32* // CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP77]], align 4 // CHECK11-NEXT: [[TMP77:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP77]], i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP78]], align 4 // CHECK11-NEXT: [[TMP78:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1 // CHECK11-NEXT: [[SUB23:%.*]] = sub nsw i32 [[TMP78]], 0
// CHECK11-NEXT: store i8* null, i8** [[TMP79]], align 4 // CHECK11-NEXT: [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
// CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2 // CHECK11-NEXT: [[SUB25:%.*]] = sub nsw i32 [[DIV24]], 1
// CHECK11-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32** // CHECK11-NEXT: store i32 [[SUB25]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP81]], align 4 // CHECK11-NEXT: [[TMP79:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2 // CHECK11-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP79]], 1
// CHECK11-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i32** // CHECK11-NEXT: [[TMP80:%.*]] = zext i32 [[ADD26]] to i64
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP83]], align 4 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP80]])
// CHECK11-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 2 // CHECK11-NEXT: [[TMP81:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP75]], i8** [[TMP76]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: store i64 [[TMP67]], i64* [[TMP84]], align 4 // CHECK11-NEXT: [[TMP82:%.*]] = icmp ne i32 [[TMP81]], 0
// CHECK11-NEXT: [[TMP85:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2 // CHECK11-NEXT: br i1 [[TMP82]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK11-NEXT: store i8* null, i8** [[TMP85]], align 4 // CHECK11: omp_offload.failed27:
// CHECK11-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i32 [[TMP57]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: [[TMP87:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK11-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK11: omp_offload.cont28:
// CHECK11-NEXT: [[TMP89:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP83:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP89]], i32* [[DOTCAPTURE_EXPR_23]], align 4 // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP83]])
// CHECK11-NEXT: [[TMP90:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP90]], 0
// CHECK11-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK11-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK11-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[TMP91:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP91]], 1
// CHECK11-NEXT: [[TMP92:%.*]] = zext i32 [[ADD28]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP92]])
// CHECK11-NEXT: [[TMP93:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP86]], i8** [[TMP87]], i64* [[TMP88]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP94:%.*]] = icmp ne i32 [[TMP93]], 0
// CHECK11-NEXT: br i1 [[TMP94]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK11: omp_offload.failed29:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i32 [[TMP65]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK11: omp_offload.cont30:
// CHECK11-NEXT: [[TMP95:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP95]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP96:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP84:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP96]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP84]])
// CHECK11-NEXT: [[TMP97:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP85:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP97]] // CHECK11-NEXT: ret i32 [[TMP85]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 222 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 126 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK11: omp_offload.failed5: // CHECK11: omp_offload.failed5:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK11: omp_offload.cont6: // CHECK11: omp_offload.cont6:
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK11-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK11-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK11-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK11: omp_offload.failed11: // CHECK11: omp_offload.failed11:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK11: omp_offload.cont12: // CHECK11: omp_offload.cont12:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK12-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK12-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED16:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS17:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS18:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS19:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[_TMP20:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES21:%.*]] = alloca [3 x i64], align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_21:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP22:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK12-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK12-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK12-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK12-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK12-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK12-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK12-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK12-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK12-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK12-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK12-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK12-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK12-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK12-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK12-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK12-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP53]])
// CHECK12-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK12-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12: omp_offload.failed14:
// CHECK12-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK12-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK12: omp_offload.cont15:
// CHECK12-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: store i32 [[TMP56]], i32* [[N_CASTED16]], align 4
// CHECK12-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK12-NEXT: [[TMP57:%.*]] = load i32, i32* [[N_CASTED16]], align 4
// CHECK12-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK12-NEXT: [[TMP58:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP61]]) // CHECK12-NEXT: [[TMP59:%.*]] = sext i32 [[TMP58]] to i64
// CHECK12-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP60:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK12-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK12-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i32*
// CHECK12-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK12-NEXT: store i32 [[TMP57]], i32* [[TMP61]], align 4
// CHECK12: omp_offload.failed15: // CHECK12-NEXT: [[TMP62:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK12-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32*
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK12-NEXT: store i32 [[TMP57]], i32* [[TMP63]], align 4
// CHECK12: omp_offload.cont16: // CHECK12-NEXT: [[TMP64:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 0
// CHECK12-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP64]], align 4
// CHECK12-NEXT: store i32 [[TMP64]], i32* [[N_CASTED17]], align 4 // CHECK12-NEXT: [[TMP65:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 1
// CHECK12-NEXT: [[TMP65:%.*]] = load i32, i32* [[N_CASTED17]], align 4 // CHECK12-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK12-NEXT: [[TMP66:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP66]], align 4
// CHECK12-NEXT: [[TMP67:%.*]] = sext i32 [[TMP66]] to i64 // CHECK12-NEXT: [[TMP67:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 1
// CHECK12-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK12-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i32*
// CHECK12-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32* // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP68]], align 4
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[TMP69]], align 4 // CHECK12-NEXT: [[TMP69:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 1
// CHECK12-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK12-NEXT: store i8* null, i8** [[TMP69]], align 4
// CHECK12-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32* // CHECK12-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[TMP71]], align 4 // CHECK12-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32**
// CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP71]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP72]], align 4 // CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 2
// CHECK12-NEXT: [[TMP73:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0 // CHECK12-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32**
// CHECK12-NEXT: store i8* null, i8** [[TMP73]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP73]], align 4
// CHECK12-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP59]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32* // CHECK12-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP75]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP74]], align 4
// CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1 // CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK12-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32* // CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP77]], align 4 // CHECK12-NEXT: [[TMP77:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP77]], i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP78]], align 4 // CHECK12-NEXT: [[TMP78:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK12-NEXT: [[TMP79:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1 // CHECK12-NEXT: [[SUB23:%.*]] = sub nsw i32 [[TMP78]], 0
// CHECK12-NEXT: store i8* null, i8** [[TMP79]], align 4 // CHECK12-NEXT: [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
// CHECK12-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2 // CHECK12-NEXT: [[SUB25:%.*]] = sub nsw i32 [[DIV24]], 1
// CHECK12-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32** // CHECK12-NEXT: store i32 [[SUB25]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP81]], align 4 // CHECK12-NEXT: [[TMP79:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2 // CHECK12-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP79]], 1
// CHECK12-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i32** // CHECK12-NEXT: [[TMP80:%.*]] = zext i32 [[ADD26]] to i64
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP83]], align 4 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP80]])
// CHECK12-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 2 // CHECK12-NEXT: [[TMP81:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP75]], i8** [[TMP76]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: store i64 [[TMP67]], i64* [[TMP84]], align 4 // CHECK12-NEXT: [[TMP82:%.*]] = icmp ne i32 [[TMP81]], 0
// CHECK12-NEXT: [[TMP85:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2 // CHECK12-NEXT: br i1 [[TMP82]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK12-NEXT: store i8* null, i8** [[TMP85]], align 4 // CHECK12: omp_offload.failed27:
// CHECK12-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i32 [[TMP57]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: [[TMP87:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK12-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK12: omp_offload.cont28:
// CHECK12-NEXT: [[TMP89:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP83:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP89]], i32* [[DOTCAPTURE_EXPR_23]], align 4 // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP83]])
// CHECK12-NEXT: [[TMP90:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP90]], 0
// CHECK12-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK12-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK12-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[TMP91:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP91]], 1
// CHECK12-NEXT: [[TMP92:%.*]] = zext i32 [[ADD28]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP92]])
// CHECK12-NEXT: [[TMP93:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP86]], i8** [[TMP87]], i64* [[TMP88]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP94:%.*]] = icmp ne i32 [[TMP93]], 0
// CHECK12-NEXT: br i1 [[TMP94]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK12: omp_offload.failed29:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i32 [[TMP65]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK12: omp_offload.cont30:
// CHECK12-NEXT: [[TMP95:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP95]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP96:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP84:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP96]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP84]])
// CHECK12-NEXT: [[TMP97:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP85:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP97]] // CHECK12-NEXT: ret i32 [[TMP85]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 222 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 // CHECK12-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 126 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK12: omp_offload.failed5: // CHECK12: omp_offload.failed5:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK12: omp_offload.cont6: // CHECK12: omp_offload.cont6:
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK12-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK12-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK12-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK12: omp_offload.failed11: // CHECK12: omp_offload.failed11:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK12: omp_offload.cont12: // CHECK12: omp_offload.cont12:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 80 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_distribute_parallel_for_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 2,439 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK9-NEXT: ret i32 [[TMP33]] // CHECK9-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 215 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 // CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK10-NEXT: ret i32 [[TMP33]] // CHECK10-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 215 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK11-NEXT: ret i32 [[TMP33]] // CHECK11-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 209 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 // CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK12-NEXT: ret i32 [[TMP33]] // CHECK12-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 1,006 Lines • ▼ Show 20 Lines
// CHECK25-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK25-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK25-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK25-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK25-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK25-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK25-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK25-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK25-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK25-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK25-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK25-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK25-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK25-NEXT: store i32 100, i32* [[N]], align 4 // CHECK25-NEXT: store i32 100, i32* [[N]], align 4
// CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK25-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK25-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK25-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK25-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK25-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK25-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK25-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK25-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK25-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK25-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK25-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK25-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK25-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK25-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK25-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK25-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK25-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK25-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK25-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK25-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK25-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK25-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK25-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK25-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK25-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK25-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK25-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK25-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK25-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK25-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK25-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK25-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK25-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK25-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK25-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK25-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK25-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK25-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK25-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK25-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK25-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK25-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK25-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK25-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK25-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK25-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK25-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK25-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK25-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK25-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK25-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK25-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK25-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK25-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK25-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK25-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK25-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK25-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK25-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK25-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK25-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK25-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK25-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK25-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK25-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK25: omp_offload.failed: // CHECK25: omp_offload.failed:
// CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK25: omp_offload.cont: // CHECK25: omp_offload.cont:
// CHECK25-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK25-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK25-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP33]]) // CHECK25-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP29]])
// CHECK25-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK25-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK25-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK25-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK25-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK25-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK25-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK25-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK25-NEXT: ret i32 [[TMP35]] // CHECK25-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162
// CHECK25-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK25-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK25-NEXT: entry: // CHECK25-NEXT: entry:
// CHECK25-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 250 Lines • ▼ Show 20 Lines
// CHECK25-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK25-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8
// CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK25-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK25-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK25-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0) // CHECK25-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0)
// CHECK25-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK25-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK25-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK25-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK25: omp_offload.failed: // CHECK25: omp_offload.failed:
// CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK25: omp_offload.cont: // CHECK25: omp_offload.cont:
// CHECK25-NEXT: ret i32 0 // CHECK25-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 173 Lines • ▼ Show 20 Lines
// CHECK26-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK26-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK26-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK26-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK26-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK26-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK26-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK26-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK26-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK26-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK26-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK26-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK26-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK26-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK26-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK26-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK26-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK26-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK26-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK26-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK26-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK26-NEXT: store i32 100, i32* [[N]], align 4 // CHECK26-NEXT: store i32 100, i32* [[N]], align 4
// CHECK26-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK26-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK26-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK26-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK26-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK26-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK26-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK26-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK26-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK26-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK26-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK26-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK26-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK26-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK26-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK26-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK26-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK26-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK26-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK26-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK26-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK26-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK26-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK26-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK26-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK26-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK26-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK26-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK26-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK26-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK26-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK26-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK26-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK26-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK26-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK26-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK26-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK26-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK26-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK26-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK26-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK26-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK26-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK26-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK26-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK26-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK26-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK26-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK26-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK26-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK26-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK26-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK26-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK26-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK26-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK26-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK26-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK26-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK26-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK26-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK26-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK26-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK26-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK26-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK26-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK26-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK26-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK26-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK26-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK26-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK26-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK26-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK26-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK26-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK26-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK26-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK26-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK26-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK26-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK26: omp_offload.failed: // CHECK26: omp_offload.failed:
// CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK26: omp_offload.cont: // CHECK26: omp_offload.cont:
// CHECK26-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK26-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK26-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP33]]) // CHECK26-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP29]])
// CHECK26-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK26-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK26-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK26-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK26-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK26-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK26-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK26-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK26-NEXT: ret i32 [[TMP35]] // CHECK26-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK26-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 // CHECK26-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162
// CHECK26-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK26-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK26-NEXT: entry: // CHECK26-NEXT: entry:
// CHECK26-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK26-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 250 Lines • ▼ Show 20 Lines
// CHECK26-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK26-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK26-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 // CHECK26-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8
// CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK26-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK26-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK26-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK26-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0) // CHECK26-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0)
// CHECK26-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK26-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK26-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK26-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK26: omp_offload.failed: // CHECK26: omp_offload.failed:
// CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK26: omp_offload.cont: // CHECK26: omp_offload.cont:
// CHECK26-NEXT: ret i32 0 // CHECK26-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 173 Lines • ▼ Show 20 Lines
// CHECK27-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK27-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK27-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK27-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK27-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK27-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK27-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK27-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK27-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK27-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK27-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK27-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK27-NEXT: store i32 100, i32* [[N]], align 4 // CHECK27-NEXT: store i32 100, i32* [[N]], align 4
// CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK27-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK27-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK27-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK27-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK27-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK27-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK27-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK27-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK27-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK27-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK27-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK27-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK27-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK27-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK27-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK27-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK27-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK27-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK27-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK27-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK27-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK27-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK27-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK27-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK27-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK27-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK27-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK27-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK27-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK27-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK27-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK27-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK27-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK27-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK27-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK27-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK27-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK27-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK27-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK27-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK27-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK27-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK27-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK27-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK27-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK27-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK27-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK27-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK27-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK27-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK27-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK27-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK27-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK27-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK27-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK27-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK27-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK27-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK27-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK27-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK27: omp_offload.failed: // CHECK27: omp_offload.failed:
// CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK27: omp_offload.cont: // CHECK27: omp_offload.cont:
// CHECK27-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK27-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK27-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP33]]) // CHECK27-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP29]])
// CHECK27-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK27-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK27-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK27-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK27-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK27-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK27-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK27-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK27-NEXT: ret i32 [[TMP35]] // CHECK27-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162
// CHECK27-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK27-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK27-NEXT: entry: // CHECK27-NEXT: entry:
// CHECK27-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 242 Lines • ▼ Show 20 Lines
// CHECK27-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK27-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4
// CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK27-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK27-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK27-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0) // CHECK27-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0)
// CHECK27-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK27-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK27-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK27-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK27: omp_offload.failed: // CHECK27: omp_offload.failed:
// CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK27: omp_offload.cont: // CHECK27: omp_offload.cont:
// CHECK27-NEXT: ret i32 0 // CHECK27-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 166 Lines • ▼ Show 20 Lines
// CHECK28-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK28-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK28-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK28-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK28-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK28-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK28-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK28-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK28-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK28-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK28-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK28-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK28-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK28-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK28-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK28-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK28-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK28-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK28-NEXT: store i32 100, i32* [[N]], align 4 // CHECK28-NEXT: store i32 100, i32* [[N]], align 4
// CHECK28-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK28-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK28-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK28-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK28-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK28-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK28-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK28-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK28-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK28-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK28-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK28-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK28-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK28-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK28-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK28-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK28-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK28-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK28-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK28-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK28-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK28-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK28-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK28-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK28-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK28-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK28-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK28-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK28-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK28-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK28-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK28-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK28-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK28-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK28-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK28-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK28-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK28-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK28-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK28-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK28-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK28-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK28-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK28-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK28-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK28-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK28-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK28-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK28-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK28-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK28-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK28-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK28-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK28-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK28-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK28-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK28-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK28-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK28-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK28-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK28-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK28-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK28-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK28-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK28-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK28-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK28-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK28-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK28-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK28-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK28-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK28-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK28-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK28-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK28-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK28-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK28: omp_offload.failed: // CHECK28: omp_offload.failed:
// CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK28: omp_offload.cont: // CHECK28: omp_offload.cont:
// CHECK28-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK28-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK28-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP33]]) // CHECK28-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP29]])
// CHECK28-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK28-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK28-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK28-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK28-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK28-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK28-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK28-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK28-NEXT: ret i32 [[TMP35]] // CHECK28-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK28-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 // CHECK28-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162
// CHECK28-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK28-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK28-NEXT: entry: // CHECK28-NEXT: entry:
// CHECK28-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK28-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 242 Lines • ▼ Show 20 Lines
// CHECK28-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK28-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK28-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 // CHECK28-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4
// CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK28-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK28-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK28-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK28-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0) // CHECK28-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0)
// CHECK28-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK28-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK28-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK28-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK28: omp_offload.failed: // CHECK28: omp_offload.failed:
// CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK28: omp_offload.cont: // CHECK28: omp_offload.cont:
// CHECK28-NEXT: ret i32 0 // CHECK28-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 160 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_distribute_parallel_for_collapse_codegen.cpp

Show First 20 Lines • Show All 976 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK9-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK9-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK9-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK9-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK9-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK9-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK9-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK9-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK9-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK9-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK9-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK9-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP54]] // CHECK9-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 298 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 182 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK10-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK10-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK10-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK10-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK10-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK10-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK10-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK10-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK10-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK10-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK10-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK10-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK10-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK10-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK10-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK10-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK10-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP54]] // CHECK10-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 298 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 182 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK11-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK11-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK11-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK11-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK11-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK11-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK11-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK11-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK11-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK11-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP53]] // CHECK11-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 298 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 176 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK12-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK12-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK12-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK12-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK12-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK12-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK12-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK12-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK12-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK12-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK12-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK12-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK12-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK12-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK12-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK12-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK12-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP53]] // CHECK12-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 298 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 159 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK12-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK12-SAME: () #[[ATTR5:[0-9]+]] { // CHECK12-SAME: () #[[ATTR5:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: call void @__tgt_register_requires(i64 1) // CHECK12-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
//
No newline at end of file

clang/test/OpenMP/teams_distribute_parallel_for_dist_schedule_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 2,397 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES24:%.*]] = alloca [4 x i64], align 8 // CHECK9-NEXT: [[_TMP23:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP25:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: store i32 10, i32* [[M]], align 4 // CHECK9-NEXT: store i32 10, i32* [[M]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK9-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK9-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK9-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK9-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK9-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 8
// CHECK9-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK9-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK9-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK9-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK9-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK9-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK9-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK9-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP52]])
// CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK9-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK9: omp_offload.failed15:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK9: omp_offload.cont16:
// CHECK9-NEXT: [[TMP55:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: [[CONV17:%.*]] = bitcast i64* [[M_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP55]], i32* [[CONV17]], align 4
// CHECK9-NEXT: [[TMP56:%.*]] = load i64, i64* [[M_CASTED]], align 8
// CHECK9-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9-NEXT: store i32 [[TMP57]], i32* [[CONV19]], align 4
// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK9-NEXT: [[TMP58:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK9-NEXT: [[TMP59:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK9-NEXT: [[TMP60:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK9-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i64*
// CHECK9-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: store i64 [[TMP56]], i64* [[TMP61]], align 8
// CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK9-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK9-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i64*
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP60]]) // CHECK9-NEXT: store i64 [[TMP56]], i64* [[TMP63]], align 8
// CHECK9-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK9-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK9-NEXT: store i8* null, i8** [[TMP64]], align 8
// CHECK9-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK9-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK9: omp_offload.failed16: // CHECK9-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i64*
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK9-NEXT: store i64 [[TMP58]], i64* [[TMP66]], align 8
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK9-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK9: omp_offload.cont17: // CHECK9-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i64*
// CHECK9-NEXT: [[TMP63:%.*]] = load i32, i32* [[M]], align 4 // CHECK9-NEXT: store i64 [[TMP58]], i64* [[TMP68]], align 8
// CHECK9-NEXT: [[CONV18:%.*]] = bitcast i64* [[M_CASTED]] to i32* // CHECK9-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK9-NEXT: store i32 [[TMP63]], i32* [[CONV18]], align 4 // CHECK9-NEXT: store i8* null, i8** [[TMP69]], align 8
// CHECK9-NEXT: [[TMP64:%.*]] = load i64, i64* [[M_CASTED]], align 8 // CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK9-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32*
// CHECK9-NEXT: store i32 [[TMP65]], i32* [[CONV20]], align 4
// CHECK9-NEXT: [[TMP66:%.*]] = load i64, i64* [[N_CASTED19]], align 8
// CHECK9-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK9-NEXT: store i64 [[TMP64]], i64* [[TMP69]], align 8
// CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK9-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64* // CHECK9-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64*
// CHECK9-NEXT: store i64 [[TMP64]], i64* [[TMP71]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP71]], align 8
// CHECK9-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 0 // CHECK9-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK9-NEXT: store i64 4, i64* [[TMP72]], align 8 // CHECK9-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i64*
// CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP73]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP73]], align 8 // CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1 // CHECK9-NEXT: store i8* null, i8** [[TMP74]], align 8
// CHECK9-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64* // CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP66]], i64* [[TMP75]], align 8 // CHECK9-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 8
// CHECK9-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64* // CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP66]], i64* [[TMP77]], align 8 // CHECK9-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32**
// CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP78]], align 8
// CHECK9-NEXT: store i64 4, i64* [[TMP78]], align 8 // CHECK9-NEXT: store i64 [[TMP59]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1 // CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK9-NEXT: store i8* null, i8** [[TMP79]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP79]], align 8
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2 // CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK9-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64* // CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP81]], align 8 // CHECK9-NEXT: [[TMP82:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2 // CHECK9-NEXT: store i32 [[TMP82]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK9-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i64* // CHECK9-NEXT: [[TMP83:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP83]], align 8 // CHECK9-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP83]], 0
// CHECK9-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 2 // CHECK9-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK9-NEXT: store i64 8, i64* [[TMP84]], align 8 // CHECK9-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK9-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2 // CHECK9-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP85]], align 8 // CHECK9-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3 // CHECK9-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP84]], 1
// CHECK9-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to i32** // CHECK9-NEXT: [[TMP85:%.*]] = zext i32 [[ADD29]] to i64
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP87]], align 8 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP85]])
// CHECK9-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3 // CHECK9-NEXT: [[TMP86:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.region_id, i32 4, i8** [[TMP80]], i8** [[TMP81]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to i32** // CHECK9-NEXT: [[TMP87:%.*]] = icmp ne i32 [[TMP86]], 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP89]], align 8 // CHECK9-NEXT: br i1 [[TMP87]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK9-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 3 // CHECK9: omp_offload.failed30:
// CHECK9-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116(i64 [[TMP56]], i64 [[TMP58]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK9-NEXT: store i8* null, i8** [[TMP91]], align 8 // CHECK9: omp_offload.cont31:
// CHECK9-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 // CHECK9-NEXT: [[TMP88:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP88]])
// CHECK9-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 0
// CHECK9-NEXT: [[TMP95:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP95]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[TMP96:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[SUB28:%.*]] = sub nsw i32 [[TMP96]], 0
// CHECK9-NEXT: [[DIV29:%.*]] = sdiv i32 [[SUB28]], 1
// CHECK9-NEXT: [[SUB30:%.*]] = sub nsw i32 [[DIV29]], 1
// CHECK9-NEXT: store i32 [[SUB30]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[ADD31:%.*]] = add nsw i32 [[TMP97]], 1
// CHECK9-NEXT: [[TMP98:%.*]] = zext i32 [[ADD31]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP98]])
// CHECK9-NEXT: [[TMP99:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.region_id, i32 4, i8** [[TMP92]], i8** [[TMP93]], i64* [[TMP94]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP100:%.*]] = icmp ne i32 [[TMP99]], 0
// CHECK9-NEXT: br i1 [[TMP100]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]]
// CHECK9: omp_offload.failed32:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116(i64 [[TMP64]], i64 [[TMP66]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT33]]
// CHECK9: omp_offload.cont33:
// CHECK9-NEXT: [[TMP101:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP101]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP102:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP89:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP102]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP89]])
// CHECK9-NEXT: [[TMP103:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP90:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP103]] // CHECK9-NEXT: ret i32 [[TMP90]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 433 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV2]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV2]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* [[CONV1]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[CONV1]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 65 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
// CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
// CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: [[CONV7:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV7:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP21]], i32* [[CONV7]], align 4 // CHECK9-NEXT: store i32 [[TMP21]], i32* [[CONV7]], align 4
// CHECK9-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i64 [[TMP1]], i32* [[TMP2]], i64 [[TMP22]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64, i32*, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i64 [[TMP1]], i32* [[TMP2]], i64 [[TMP22]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
// CHECK9-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 25 Lines
// CHECK9-NEXT: [[TMP34:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP34:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP34]], align 4 // CHECK9-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP34]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP35]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP35]])
// CHECK9-NEXT: br label [[OMP_PRECOND_END]] // CHECK9-NEXT: br label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.end: // CHECK9: omp.precond.end:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 123 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK9: omp_offload.failed5: // CHECK9: omp_offload.failed5:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK9: omp_offload.cont6: // CHECK9: omp_offload.cont6:
// CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[M_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[M_CASTED]] to i32*
Show All 13 Lines
// CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]** // CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 8
// CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 // CHECK9-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// CHECK9-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK9-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK9: omp_offload.failed11: // CHECK9: omp_offload.failed11:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94(i64 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94(i64 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK9: omp_offload.cont12: // CHECK9: omp_offload.cont12:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 66 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 77 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[M_ADDR]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[M_ADDR]] to i32*
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4 // CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP3]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP3]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 39 Lines
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4 // CHECK9-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4
// CHECK9-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 19 Lines
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK10-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES24:%.*]] = alloca [4 x i64], align 8 // CHECK10-NEXT: [[_TMP23:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP25:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: store i32 10, i32* [[M]], align 4 // CHECK10-NEXT: store i32 10, i32* [[M]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK10-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK10-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK10-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK10-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK10-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK10-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 8
// CHECK10-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK10-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK10-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK10-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK10-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK10-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK10-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK10-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK10-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK10-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK10-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK10-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK10-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK10-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK10-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP52]])
// CHECK10-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK10-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK10: omp_offload.failed15:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK10: omp_offload.cont16:
// CHECK10-NEXT: [[TMP55:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: [[CONV17:%.*]] = bitcast i64* [[M_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP55]], i32* [[CONV17]], align 4
// CHECK10-NEXT: [[TMP56:%.*]] = load i64, i64* [[M_CASTED]], align 8
// CHECK10-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10-NEXT: store i32 [[TMP57]], i32* [[CONV19]], align 4
// CHECK10-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK10-NEXT: [[TMP58:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK10-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK10-NEXT: [[TMP59:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK10-NEXT: [[TMP60:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK10-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i64*
// CHECK10-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: store i64 [[TMP56]], i64* [[TMP61]], align 8
// CHECK10-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK10-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK10-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i64*
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP60]]) // CHECK10-NEXT: store i64 [[TMP56]], i64* [[TMP63]], align 8
// CHECK10-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK10-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK10-NEXT: store i8* null, i8** [[TMP64]], align 8
// CHECK10-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK10-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK10: omp_offload.failed16: // CHECK10-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i64*
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK10-NEXT: store i64 [[TMP58]], i64* [[TMP66]], align 8
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK10-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK10: omp_offload.cont17: // CHECK10-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i64*
// CHECK10-NEXT: [[TMP63:%.*]] = load i32, i32* [[M]], align 4 // CHECK10-NEXT: store i64 [[TMP58]], i64* [[TMP68]], align 8
// CHECK10-NEXT: [[CONV18:%.*]] = bitcast i64* [[M_CASTED]] to i32* // CHECK10-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK10-NEXT: store i32 [[TMP63]], i32* [[CONV18]], align 4 // CHECK10-NEXT: store i8* null, i8** [[TMP69]], align 8
// CHECK10-NEXT: [[TMP64:%.*]] = load i64, i64* [[M_CASTED]], align 8 // CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK10-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32*
// CHECK10-NEXT: store i32 [[TMP65]], i32* [[CONV20]], align 4
// CHECK10-NEXT: [[TMP66:%.*]] = load i64, i64* [[N_CASTED19]], align 8
// CHECK10-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK10-NEXT: store i64 [[TMP64]], i64* [[TMP69]], align 8
// CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK10-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64* // CHECK10-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64*
// CHECK10-NEXT: store i64 [[TMP64]], i64* [[TMP71]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP71]], align 8
// CHECK10-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 0 // CHECK10-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK10-NEXT: store i64 4, i64* [[TMP72]], align 8 // CHECK10-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i64*
// CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP73]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP73]], align 8 // CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1 // CHECK10-NEXT: store i8* null, i8** [[TMP74]], align 8
// CHECK10-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64* // CHECK10-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK10-NEXT: store i64 [[TMP66]], i64* [[TMP75]], align 8 // CHECK10-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK10-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 8
// CHECK10-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64* // CHECK10-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK10-NEXT: store i64 [[TMP66]], i64* [[TMP77]], align 8 // CHECK10-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32**
// CHECK10-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP78]], align 8
// CHECK10-NEXT: store i64 4, i64* [[TMP78]], align 8 // CHECK10-NEXT: store i64 [[TMP59]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK10-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1 // CHECK10-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK10-NEXT: store i8* null, i8** [[TMP79]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP79]], align 8
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2 // CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK10-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64* // CHECK10-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP81]], align 8 // CHECK10-NEXT: [[TMP82:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2 // CHECK10-NEXT: store i32 [[TMP82]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK10-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i64* // CHECK10-NEXT: [[TMP83:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP83]], align 8 // CHECK10-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP83]], 0
// CHECK10-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 2 // CHECK10-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK10-NEXT: store i64 8, i64* [[TMP84]], align 8 // CHECK10-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK10-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2 // CHECK10-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP85]], align 8 // CHECK10-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3 // CHECK10-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP84]], 1
// CHECK10-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to i32** // CHECK10-NEXT: [[TMP85:%.*]] = zext i32 [[ADD29]] to i64
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP87]], align 8 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP85]])
// CHECK10-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3 // CHECK10-NEXT: [[TMP86:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.region_id, i32 4, i8** [[TMP80]], i8** [[TMP81]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to i32** // CHECK10-NEXT: [[TMP87:%.*]] = icmp ne i32 [[TMP86]], 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP89]], align 8 // CHECK10-NEXT: br i1 [[TMP87]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK10-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 3 // CHECK10: omp_offload.failed30:
// CHECK10-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116(i64 [[TMP56]], i64 [[TMP58]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK10-NEXT: store i8* null, i8** [[TMP91]], align 8 // CHECK10: omp_offload.cont31:
// CHECK10-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 // CHECK10-NEXT: [[TMP88:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP88]])
// CHECK10-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 0
// CHECK10-NEXT: [[TMP95:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP95]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[TMP96:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[SUB28:%.*]] = sub nsw i32 [[TMP96]], 0
// CHECK10-NEXT: [[DIV29:%.*]] = sdiv i32 [[SUB28]], 1
// CHECK10-NEXT: [[SUB30:%.*]] = sub nsw i32 [[DIV29]], 1
// CHECK10-NEXT: store i32 [[SUB30]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[ADD31:%.*]] = add nsw i32 [[TMP97]], 1
// CHECK10-NEXT: [[TMP98:%.*]] = zext i32 [[ADD31]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP98]])
// CHECK10-NEXT: [[TMP99:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.region_id, i32 4, i8** [[TMP92]], i8** [[TMP93]], i64* [[TMP94]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP100:%.*]] = icmp ne i32 [[TMP99]], 0
// CHECK10-NEXT: br i1 [[TMP100]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]]
// CHECK10: omp_offload.failed32:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116(i64 [[TMP64]], i64 [[TMP66]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT33]]
// CHECK10: omp_offload.cont33:
// CHECK10-NEXT: [[TMP101:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP101]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP102:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP89:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP102]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP89]])
// CHECK10-NEXT: [[TMP103:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP90:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP103]] // CHECK10-NEXT: ret i32 [[TMP90]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 433 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV2]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV2]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* [[CONV1]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[CONV1]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 65 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 // CHECK10-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
// CHECK10-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 // CHECK10-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
// CHECK10-NEXT: [[TMP21:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP21:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: [[CONV7:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV7:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP21]], i32* [[CONV7]], align 4 // CHECK10-NEXT: store i32 [[TMP21]], i32* [[CONV7]], align 4
// CHECK10-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i64 [[TMP1]], i32* [[TMP2]], i64 [[TMP22]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64, i32*, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i64 [[TMP1]], i32* [[TMP2]], i64 [[TMP22]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] // CHECK10-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
// CHECK10-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 25 Lines
// CHECK10-NEXT: [[TMP34:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP34:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP34]], align 4 // CHECK10-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP34]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP35]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP35]])
// CHECK10-NEXT: br label [[OMP_PRECOND_END]] // CHECK10-NEXT: br label [[OMP_PRECOND_END]]
// CHECK10: omp.precond.end: // CHECK10: omp.precond.end:
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 123 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK10: omp_offload.failed5: // CHECK10: omp_offload.failed5:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK10: omp_offload.cont6: // CHECK10: omp_offload.cont6:
// CHECK10-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK10-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[M_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[M_CASTED]] to i32*
Show All 13 Lines
// CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]** // CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 8
// CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1 // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 // CHECK10-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// CHECK10-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK10-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK10: omp_offload.failed11: // CHECK10: omp_offload.failed11:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94(i64 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94(i64 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK10: omp_offload.cont12: // CHECK10: omp_offload.cont12:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 66 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 77 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[M_ADDR]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[M_ADDR]] to i32*
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4 // CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP3]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP3]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 39 Lines
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4 // CHECK10-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4
// CHECK10-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 19 Lines
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK11-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED16:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS17:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS19:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES21:%.*]] = alloca [4 x i64], align 4 // CHECK11-NEXT: [[_TMP20:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP22:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_21:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: store i32 10, i32* [[M]], align 4 // CHECK11-NEXT: store i32 10, i32* [[M]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK11-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK11-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK11-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK11-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK11-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK11-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK11-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK11-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK11-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK11-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP53]])
// CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK11-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK11: omp_offload.failed14:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK11: omp_offload.cont15:
// CHECK11-NEXT: [[TMP56:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP56]], i32* [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP57:%.*]] = load i32, i32* [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11-NEXT: store i32 [[TMP58]], i32* [[N_CASTED16]], align 4
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[N_CASTED16]], align 4
// CHECK11-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK11-NEXT: [[TMP60:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK11-NEXT: [[TMP61:%.*]] = sext i32 [[TMP60]] to i64
// CHECK11-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK11-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32*
// CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: store i32 [[TMP57]], i32* [[TMP63]], align 4
// CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK11-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK11-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK11-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32*
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP61]]) // CHECK11-NEXT: store i32 [[TMP57]], i32* [[TMP65]], align 4
// CHECK11-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 0
// CHECK11-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK11-NEXT: store i8* null, i8** [[TMP66]], align 4
// CHECK11-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 1
// CHECK11: omp_offload.failed15: // CHECK11-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i32*
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK11-NEXT: store i32 [[TMP59]], i32* [[TMP68]], align 4
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK11-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 1
// CHECK11: omp_offload.cont16: // CHECK11-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i32*
// CHECK11-NEXT: [[TMP64:%.*]] = load i32, i32* [[M]], align 4 // CHECK11-NEXT: store i32 [[TMP59]], i32* [[TMP70]], align 4
// CHECK11-NEXT: store i32 [[TMP64]], i32* [[M_CASTED]], align 4 // CHECK11-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 1
// CHECK11-NEXT: [[TMP65:%.*]] = load i32, i32* [[M_CASTED]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP71]], align 4
// CHECK11-NEXT: [[TMP66:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP66]], i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: [[TMP67:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: [[TMP68:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP69:%.*]] = sext i32 [[TMP68]] to i64
// CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK11-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[TMP71]], align 4
// CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK11-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32* // CHECK11-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32*
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[TMP73]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP73]], align 4
// CHECK11-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK11-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP74]], align 4 // CHECK11-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32*
// CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP75]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP75]], align 4 // CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 2
// CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP76]], align 4
// CHECK11-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32* // CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP67]], i32* [[TMP77]], align 4 // CHECK11-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32**
// CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP78]], align 4
// CHECK11-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32* // CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP67]], i32* [[TMP79]], align 4 // CHECK11-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32**
// CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP80]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP80]], align 4 // CHECK11-NEXT: store i64 [[TMP61]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1 // CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP81]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP81]], align 4
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2 // CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK11-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i32* // CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP83]], align 4 // CHECK11-NEXT: [[TMP84:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[TMP84]], i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK11-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to i32* // CHECK11-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP85]], align 4 // CHECK11-NEXT: [[SUB23:%.*]] = sub nsw i32 [[TMP85]], 0
// CHECK11-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 2 // CHECK11-NEXT: [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
// CHECK11-NEXT: store i64 4, i64* [[TMP86]], align 4 // CHECK11-NEXT: [[SUB25:%.*]] = sub nsw i32 [[DIV24]], 1
// CHECK11-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[SUB25]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP87]], align 4 // CHECK11-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3 // CHECK11-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP86]], 1
// CHECK11-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to i32** // CHECK11-NEXT: [[TMP87:%.*]] = zext i32 [[ADD26]] to i64
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP89]], align 4 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP87]])
// CHECK11-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3 // CHECK11-NEXT: [[TMP88:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.region_id, i32 4, i8** [[TMP82]], i8** [[TMP83]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to i32** // CHECK11-NEXT: [[TMP89:%.*]] = icmp ne i32 [[TMP88]], 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP91]], align 4 // CHECK11-NEXT: br i1 [[TMP89]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK11-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 3 // CHECK11: omp_offload.failed27:
// CHECK11-NEXT: store i64 [[TMP69]], i64* [[TMP92]], align 4 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116(i32 [[TMP57]], i32 [[TMP59]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK11-NEXT: store i8* null, i8** [[TMP93]], align 4 // CHECK11: omp_offload.cont28:
// CHECK11-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK11-NEXT: [[TMP90:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP90]])
// CHECK11-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0
// CHECK11-NEXT: [[TMP97:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP97]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP98]], 0
// CHECK11-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK11-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK11-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP99]], 1
// CHECK11-NEXT: [[TMP100:%.*]] = zext i32 [[ADD28]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP100]])
// CHECK11-NEXT: [[TMP101:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.region_id, i32 4, i8** [[TMP94]], i8** [[TMP95]], i64* [[TMP96]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP102:%.*]] = icmp ne i32 [[TMP101]], 0
// CHECK11-NEXT: br i1 [[TMP102]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK11: omp_offload.failed29:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116(i32 [[TMP65]], i32 [[TMP67]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK11: omp_offload.cont30:
// CHECK11-NEXT: [[TMP103:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP103]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP104:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP91:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP104]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP91]])
// CHECK11-NEXT: [[TMP105:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP92:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP105]] // CHECK11-NEXT: ret i32 [[TMP92]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 418 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[M_ADDR]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[M_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]] // CHECK11-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
// CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP19]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP19]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32 [[TMP1]], i32* [[TMP2]], i32 [[TMP20]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32, i32*, i32)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32 [[TMP1]], i32* [[TMP2]], i32 [[TMP20]])
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
// CHECK11-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 25 Lines
// CHECK11-NEXT: [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK11-NEXT: [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP33]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP33]])
// CHECK11-NEXT: br label [[OMP_PRECOND_END]] // CHECK11-NEXT: br label [[OMP_PRECOND_END]]
// CHECK11: omp.precond.end: // CHECK11: omp.precond.end:
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 119 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK11: omp_offload.failed5: // CHECK11: omp_offload.failed5:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK11: omp_offload.cont6: // CHECK11: omp_offload.cont6:
// CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP18]], i32* [[M_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP18]], i32* [[M_CASTED]], align 4
Show All 12 Lines
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK11-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]** // CHECK11-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP29]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 // CHECK11-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// CHECK11-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK11-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK11: omp_offload.failed11: // CHECK11: omp_offload.failed11:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94(i32 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94(i32 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK11: omp_offload.cont12: // CHECK11: omp_offload.cont12:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]])
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]])
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 72 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: store i32 [[M]], i32* [[M_ADDR]], align 4 // CHECK11-NEXT: store i32 [[M]], i32* [[M_ADDR]], align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[M_ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[M_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP3]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP3]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 35 Lines
// CHECK11-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10 // CHECK11-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]])
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 19 Lines
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 80 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK12-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED16:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS17:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS19:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES21:%.*]] = alloca [4 x i64], align 4 // CHECK12-NEXT: [[_TMP20:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP22:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_21:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: store i32 10, i32* [[M]], align 4 // CHECK12-NEXT: store i32 10, i32* [[M]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK12-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK12-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK12-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK12-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK12-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK12-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK12-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK12-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK12-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK12-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK12-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK12-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK12-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK12-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK12-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK12-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP53]])
// CHECK12-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK12-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK12: omp_offload.failed14:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK12: omp_offload.cont15:
// CHECK12-NEXT: [[TMP56:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP56]], i32* [[M_CASTED]], align 4
// CHECK12-NEXT: [[TMP57:%.*]] = load i32, i32* [[M_CASTED]], align 4
// CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12-NEXT: store i32 [[TMP58]], i32* [[N_CASTED16]], align 4
// CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[N_CASTED16]], align 4
// CHECK12-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK12-NEXT: [[TMP60:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK12-NEXT: [[TMP61:%.*]] = sext i32 [[TMP60]] to i64
// CHECK12-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK12-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK12-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32*
// CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: store i32 [[TMP57]], i32* [[TMP63]], align 4
// CHECK12-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK12-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK12-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK12-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32*
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP61]]) // CHECK12-NEXT: store i32 [[TMP57]], i32* [[TMP65]], align 4
// CHECK12-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 0
// CHECK12-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK12-NEXT: store i8* null, i8** [[TMP66]], align 4
// CHECK12-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK12-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 1
// CHECK12: omp_offload.failed15: // CHECK12-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i32*
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK12-NEXT: store i32 [[TMP59]], i32* [[TMP68]], align 4
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK12-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 1
// CHECK12: omp_offload.cont16: // CHECK12-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i32*
// CHECK12-NEXT: [[TMP64:%.*]] = load i32, i32* [[M]], align 4 // CHECK12-NEXT: store i32 [[TMP59]], i32* [[TMP70]], align 4
// CHECK12-NEXT: store i32 [[TMP64]], i32* [[M_CASTED]], align 4 // CHECK12-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 1
// CHECK12-NEXT: [[TMP65:%.*]] = load i32, i32* [[M_CASTED]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP71]], align 4
// CHECK12-NEXT: [[TMP66:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP66]], i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: [[TMP67:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: [[TMP68:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP69:%.*]] = sext i32 [[TMP68]] to i64
// CHECK12-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK12-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[TMP71]], align 4
// CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK12-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32* // CHECK12-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32*
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[TMP73]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP73]], align 4
// CHECK12-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK12-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP74]], align 4 // CHECK12-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32*
// CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP75]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP75]], align 4 // CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 2
// CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP76]], align 4
// CHECK12-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32* // CHECK12-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP67]], i32* [[TMP77]], align 4 // CHECK12-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32**
// CHECK12-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP78]], align 4
// CHECK12-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32* // CHECK12-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP67]], i32* [[TMP79]], align 4 // CHECK12-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32**
// CHECK12-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP80]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP80]], align 4 // CHECK12-NEXT: store i64 [[TMP61]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK12-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1 // CHECK12-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP81]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP81]], align 4
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2 // CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK12-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i32* // CHECK12-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP83]], align 4 // CHECK12-NEXT: [[TMP84:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[TMP84]], i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK12-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to i32* // CHECK12-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP85]], align 4 // CHECK12-NEXT: [[SUB23:%.*]] = sub nsw i32 [[TMP85]], 0
// CHECK12-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 2 // CHECK12-NEXT: [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
// CHECK12-NEXT: store i64 4, i64* [[TMP86]], align 4 // CHECK12-NEXT: [[SUB25:%.*]] = sub nsw i32 [[DIV24]], 1
// CHECK12-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[SUB25]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP87]], align 4 // CHECK12-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3 // CHECK12-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP86]], 1
// CHECK12-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to i32** // CHECK12-NEXT: [[TMP87:%.*]] = zext i32 [[ADD26]] to i64
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP89]], align 4 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP87]])
// CHECK12-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3 // CHECK12-NEXT: [[TMP88:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.region_id, i32 4, i8** [[TMP82]], i8** [[TMP83]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to i32** // CHECK12-NEXT: [[TMP89:%.*]] = icmp ne i32 [[TMP88]], 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP91]], align 4 // CHECK12-NEXT: br i1 [[TMP89]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK12-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 3 // CHECK12: omp_offload.failed27:
// CHECK12-NEXT: store i64 [[TMP69]], i64* [[TMP92]], align 4 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116(i32 [[TMP57]], i32 [[TMP59]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK12-NEXT: store i8* null, i8** [[TMP93]], align 4 // CHECK12: omp_offload.cont28:
// CHECK12-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK12-NEXT: [[TMP90:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP90]])
// CHECK12-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0
// CHECK12-NEXT: [[TMP97:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP97]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP98]], 0
// CHECK12-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK12-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK12-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP99]], 1
// CHECK12-NEXT: [[TMP100:%.*]] = zext i32 [[ADD28]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP100]])
// CHECK12-NEXT: [[TMP101:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.region_id, i32 4, i8** [[TMP94]], i8** [[TMP95]], i64* [[TMP96]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP102:%.*]] = icmp ne i32 [[TMP101]], 0
// CHECK12-NEXT: br i1 [[TMP102]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK12: omp_offload.failed29:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116(i32 [[TMP65]], i32 [[TMP67]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK12: omp_offload.cont30:
// CHECK12-NEXT: [[TMP103:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP103]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP104:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP91:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP104]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP91]])
// CHECK12-NEXT: [[TMP105:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP92:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP105]] // CHECK12-NEXT: ret i32 [[TMP92]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 418 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 // CHECK12-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[M_ADDR]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[M_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]] // CHECK12-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
// CHECK12-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP19]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP19]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32 [[TMP1]], i32* [[TMP2]], i32 [[TMP20]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32, i32*, i32)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32 [[TMP1]], i32* [[TMP2]], i32 [[TMP20]])
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK12-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] // CHECK12-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
// CHECK12-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 25 Lines
// CHECK12-NEXT: [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK12-NEXT: [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4 // CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP33]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP33]])
// CHECK12-NEXT: br label [[OMP_PRECOND_END]] // CHECK12-NEXT: br label [[OMP_PRECOND_END]]
// CHECK12: omp.precond.end: // CHECK12: omp.precond.end:
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 119 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK12: omp_offload.failed5: // CHECK12: omp_offload.failed5:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK12: omp_offload.cont6: // CHECK12: omp_offload.cont6:
// CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP18]], i32* [[M_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP18]], i32* [[M_CASTED]], align 4
Show All 12 Lines
// CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK12-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]** // CHECK12-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1 // CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP29]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 // CHECK12-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// CHECK12-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK12-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK12: omp_offload.failed11: // CHECK12: omp_offload.failed11:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94(i32 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94(i32 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK12: omp_offload.cont12: // CHECK12: omp_offload.cont12:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]])
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]])
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 72 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: store i32 [[M]], i32* [[M_ADDR]], align 4 // CHECK12-NEXT: store i32 [[M]], i32* [[M_ADDR]], align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[M_ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[M_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP3]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP3]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 35 Lines
// CHECK12-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10 // CHECK12-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]])
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
Show All 19 Lines
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 73 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_distribute_parallel_for_schedule_codegen.cpp

This file has a very large number of changes (5,360 lines). Show File Contents

clang/test/OpenMP/teams_distribute_parallel_for_simd_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 3,271 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[I_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
Show All 9 Lines
// CHECK9-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64* // CHECK9-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP11]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP11]], align 8
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 8, i64* [[TMP12]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32**
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32** // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP15]], align 8 // CHECK9-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32**
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32** // CHECK9-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 1), align 8
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP18]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64*
// CHECK9-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP21]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP23]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i64*
// CHECK9-NEXT: store i64 4, i64* [[TMP24]], align 8 // CHECK9-NEXT: store i64 [[TMP6]], i64* [[TMP24]], align 8
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i8* null, i8** [[TMP25]], align 8 // CHECK9-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: store i64 [[TMP6]], i64* [[TMP26]], align 8
// CHECK9-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store i64 [[TMP6]], i64* [[TMP27]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP27]], align 8
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP6]], i64* [[TMP29]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK9-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store i64 4, i64* [[TMP30]], align 8 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK9-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK9-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK9-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK9-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK9-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK9-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74(i64 [[TMP1]], i32* [[VLA]], i64 [[TMP4]], i64 [[TMP6]]) #[[ATTR5:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74(i64 [[TMP1]], i32* [[VLA]], i64 [[TMP4]], i64 [[TMP6]]) #[[ATTR5:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0
// CHECK9-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK9-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK9-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK9-NEXT: ret i32 [[TMP41]] // CHECK9-NEXT: ret i32 [[TMP36]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74
// CHECK9-SAME: (i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[I:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[I:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[I_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[I_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 265 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[I_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
Show All 9 Lines
// CHECK10-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64* // CHECK10-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP11]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP11]], align 8
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 8, i64* [[TMP12]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32**
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32** // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP15]], align 8 // CHECK10-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32**
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP16]], align 8
// CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32** // CHECK10-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 1), align 8
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP18]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64*
// CHECK10-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP21]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK10-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP23]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i64*
// CHECK10-NEXT: store i64 4, i64* [[TMP24]], align 8 // CHECK10-NEXT: store i64 [[TMP6]], i64* [[TMP24]], align 8
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK10-NEXT: store i8* null, i8** [[TMP25]], align 8 // CHECK10-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK10-NEXT: store i64 [[TMP6]], i64* [[TMP26]], align 8
// CHECK10-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK10-NEXT: store i64 [[TMP6]], i64* [[TMP27]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP27]], align 8
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP6]], i64* [[TMP29]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK10-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: store i64 4, i64* [[TMP30]], align 8 // CHECK10-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK10-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK10-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK10-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK10-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK10-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK10-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK10-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK10-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74(i64 [[TMP1]], i32* [[VLA]], i64 [[TMP4]], i64 [[TMP6]]) #[[ATTR5:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74(i64 [[TMP1]], i32* [[VLA]], i64 [[TMP4]], i64 [[TMP6]]) #[[ATTR5:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 // CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0
// CHECK10-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK10-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK10-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK10-NEXT: ret i32 [[TMP41]] // CHECK10-NEXT: ret i32 [[TMP36]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74
// CHECK10-SAME: (i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[I:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[I:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[I_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[I_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 265 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[I_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[I_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[I]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[I]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[I_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[I_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[I_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[I_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP4]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP4]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64 // CHECK11-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32* // CHECK11-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP11]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP11]], align 4
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP12]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32**
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32** // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP15]], align 4 // CHECK11-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32**
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP16]], align 4
// CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32** // CHECK11-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 1), align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK11-NEXT: store i64 [[TMP7]], i64* [[TMP18]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK11-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32* // CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP21]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP21]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP23]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK11-NEXT: store i64 4, i64* [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP25]], align 4 // CHECK11-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP26]], align 4
// CHECK11-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32* // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP27]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP29]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP30]], align 4 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK11-NEXT: store i8* null, i8** [[TMP31]], align 4
// CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK11-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK11-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK11-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK11-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74(i32 [[TMP0]], i32* [[VLA]], i32 [[TMP3]], i32 [[TMP5]]) #[[ATTR5:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74(i32 [[TMP0]], i32* [[VLA]], i32 [[TMP3]], i32 [[TMP5]]) #[[ATTR5:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0
// CHECK11-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK11-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK11-NEXT: ret i32 [[TMP41]] // CHECK11-NEXT: ret i32 [[TMP36]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74
// CHECK11-SAME: (i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[I:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[I:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 258 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[I_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[I_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[I]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[I]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[I_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[I_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[I_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[I_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP4]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP4]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64 // CHECK12-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32* // CHECK12-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP11]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP11]], align 4
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP12]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32**
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32** // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP15]], align 4 // CHECK12-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32**
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP16]], align 4
// CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32** // CHECK12-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 1), align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK12-NEXT: store i64 [[TMP7]], i64* [[TMP18]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK12-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32* // CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP21]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP21]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP23]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK12-NEXT: store i64 4, i64* [[TMP24]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP24]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP25]], align 4 // CHECK12-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP26]], align 4
// CHECK12-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32* // CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP27]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP29]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP30]], align 4 // CHECK12-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK12-NEXT: store i8* null, i8** [[TMP31]], align 4
// CHECK12-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK12-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK12-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK12-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK12-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74(i32 [[TMP0]], i32* [[VLA]], i32 [[TMP3]], i32 [[TMP5]]) #[[ATTR5:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74(i32 [[TMP0]], i32* [[VLA]], i32 [[TMP3]], i32 [[TMP5]]) #[[ATTR5:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 // CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0
// CHECK12-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK12-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK12-NEXT: ret i32 [[TMP41]] // CHECK12-NEXT: ret i32 [[TMP36]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l74
// CHECK12-SAME: (i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[I:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[I:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 1,845 Lines • ▼ Show 20 Lines
// CHECK25-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK25-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK25-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK25-NEXT: [[I_CASTED:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[I_CASTED:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 // CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
// CHECK25-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 8
// CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK25-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK25-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK25-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK25-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK25-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK25-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK25-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK25-NEXT: store i32 100, i32* [[N]], align 4 // CHECK25-NEXT: store i32 100, i32* [[N]], align 4
// CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
Show All 12 Lines
// CHECK25-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK25-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK25-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK25-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK25-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK25-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP9]], align 8 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP9]], align 8
// CHECK25-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64* // CHECK25-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64*
// CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP11]], align 8 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP11]], align 8
// CHECK25-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK25-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK25-NEXT: store i64 8, i64* [[TMP12]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP12]], align 8
// CHECK25-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK25-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK25-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK25-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32**
// CHECK25-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP14]], align 8
// CHECK25-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32** // CHECK25-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP15]], align 8 // CHECK25-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32**
// CHECK25-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP16]], align 8
// CHECK25-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32** // CHECK25-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 1), align 8
// CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8 // CHECK25-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK25-NEXT: store i8* null, i8** [[TMP17]], align 8
// CHECK25-NEXT: store i64 [[TMP7]], i64* [[TMP18]], align 8 // CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK25-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64*
// CHECK25-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP19]], align 8
// CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK25-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK25-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP21]], align 8 // CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP21]], align 8
// CHECK25-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK25-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK25-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK25-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP23]], align 8 // CHECK25-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK25-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK25-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i64*
// CHECK25-NEXT: store i64 4, i64* [[TMP24]], align 8 // CHECK25-NEXT: store i64 [[TMP6]], i64* [[TMP24]], align 8
// CHECK25-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK25-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK25-NEXT: store i8* null, i8** [[TMP25]], align 8 // CHECK25-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK25-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK25-NEXT: store i64 [[TMP6]], i64* [[TMP26]], align 8
// CHECK25-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK25-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK25-NEXT: store i64 [[TMP6]], i64* [[TMP27]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP27]], align 8
// CHECK25-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK25-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK25-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK25-NEXT: store i64 [[TMP6]], i64* [[TMP29]], align 8 // CHECK25-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK25-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK25-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: store i64 4, i64* [[TMP30]], align 8 // CHECK25-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK25-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK25-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK25-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK25-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK25-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK25-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK25-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK25-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK25-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK25-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK25-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK25-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK25-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK25-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK25-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK25-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK25-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK25-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK25: omp_offload.failed: // CHECK25: omp_offload.failed:
// CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166(i64 [[TMP1]], i32* [[VLA]], i64 [[TMP4]], i64 [[TMP6]]) #[[ATTR5:[0-9]+]] // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166(i64 [[TMP1]], i32* [[VLA]], i64 [[TMP4]], i64 [[TMP6]]) #[[ATTR5:[0-9]+]]
// CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK25: omp_offload.cont: // CHECK25: omp_offload.cont:
// CHECK25-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK25-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK25-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP41]]) // CHECK25-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP36]])
// CHECK25-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK25-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK25-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK25-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK25-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK25-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK25-NEXT: [[TMP43:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK25-NEXT: [[TMP38:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK25-NEXT: ret i32 [[TMP43]] // CHECK25-NEXT: ret i32 [[TMP38]]
// //
// //
// CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166
// CHECK25-SAME: (i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[I:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK25-SAME: (i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[I:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK25-NEXT: entry: // CHECK25-NEXT: entry:
// CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK25-NEXT: [[I_ADDR:%.*]] = alloca i64, align 8 // CHECK25-NEXT: [[I_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 298 Lines • ▼ Show 20 Lines
// CHECK25-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK25-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8
// CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK25-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK25-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK25-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK25-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i64 10) // CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i64 10)
// CHECK25-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0) // CHECK25-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0)
// CHECK25-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK25-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK25-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK25-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK25: omp_offload.failed: // CHECK25: omp_offload.failed:
// CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR5]] // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR5]]
// CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK25: omp_offload.cont: // CHECK25: omp_offload.cont:
// CHECK25-NEXT: ret i32 0 // CHECK25-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 189 Lines • ▼ Show 20 Lines
// CHECK26-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK26-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK26-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK26-NEXT: [[I_CASTED:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[I_CASTED:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK26-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK26-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK26-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK26-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 // CHECK26-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
// CHECK26-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 8
// CHECK26-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK26-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK26-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK26-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK26-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK26-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK26-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK26-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK26-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK26-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK26-NEXT: store i32 100, i32* [[N]], align 4 // CHECK26-NEXT: store i32 100, i32* [[N]], align 4
// CHECK26-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK26-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
Show All 12 Lines
// CHECK26-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK26-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK26-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK26-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK26-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK26-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP9]], align 8 // CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP9]], align 8
// CHECK26-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64* // CHECK26-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64*
// CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP11]], align 8 // CHECK26-NEXT: store i64 [[TMP1]], i64* [[TMP11]], align 8
// CHECK26-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK26-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK26-NEXT: store i64 8, i64* [[TMP12]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP12]], align 8
// CHECK26-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK26-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK26-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK26-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32**
// CHECK26-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP14]], align 8
// CHECK26-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32** // CHECK26-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP15]], align 8 // CHECK26-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32**
// CHECK26-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP16]], align 8
// CHECK26-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32** // CHECK26-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 1), align 8
// CHECK26-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8 // CHECK26-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK26-NEXT: store i8* null, i8** [[TMP17]], align 8
// CHECK26-NEXT: store i64 [[TMP7]], i64* [[TMP18]], align 8 // CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK26-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK26-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64*
// CHECK26-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP19]], align 8
// CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK26-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK26-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP21]], align 8 // CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP21]], align 8
// CHECK26-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK26-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK26-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK26-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK26-NEXT: store i64 [[TMP4]], i64* [[TMP23]], align 8 // CHECK26-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK26-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK26-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i64*
// CHECK26-NEXT: store i64 4, i64* [[TMP24]], align 8 // CHECK26-NEXT: store i64 [[TMP6]], i64* [[TMP24]], align 8
// CHECK26-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK26-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK26-NEXT: store i8* null, i8** [[TMP25]], align 8 // CHECK26-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK26-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK26-NEXT: store i64 [[TMP6]], i64* [[TMP26]], align 8
// CHECK26-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK26-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK26-NEXT: store i64 [[TMP6]], i64* [[TMP27]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP27]], align 8
// CHECK26-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK26-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK26-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK26-NEXT: store i64 [[TMP6]], i64* [[TMP29]], align 8 // CHECK26-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK26-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK26-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: store i64 4, i64* [[TMP30]], align 8 // CHECK26-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK26-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK26-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK26-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK26-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK26-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK26-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK26-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK26-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK26-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK26-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK26-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK26-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK26-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK26-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK26-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK26-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK26-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK26-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK26-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK26-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK26-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK26-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK26-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK26-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK26: omp_offload.failed: // CHECK26: omp_offload.failed:
// CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166(i64 [[TMP1]], i32* [[VLA]], i64 [[TMP4]], i64 [[TMP6]]) #[[ATTR5:[0-9]+]] // CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166(i64 [[TMP1]], i32* [[VLA]], i64 [[TMP4]], i64 [[TMP6]]) #[[ATTR5:[0-9]+]]
// CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK26: omp_offload.cont: // CHECK26: omp_offload.cont:
// CHECK26-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK26-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK26-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP41]]) // CHECK26-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP36]])
// CHECK26-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK26-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK26-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK26-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK26-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK26-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK26-NEXT: [[TMP43:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK26-NEXT: [[TMP38:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK26-NEXT: ret i32 [[TMP43]] // CHECK26-NEXT: ret i32 [[TMP38]]
// //
// //
// CHECK26-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166 // CHECK26-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166
// CHECK26-SAME: (i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[I:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK26-SAME: (i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[I:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK26-NEXT: entry: // CHECK26-NEXT: entry:
// CHECK26-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK26-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK26-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK26-NEXT: [[I_ADDR:%.*]] = alloca i64, align 8 // CHECK26-NEXT: [[I_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 298 Lines • ▼ Show 20 Lines
// CHECK26-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK26-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK26-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 // CHECK26-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8
// CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK26-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK26-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK26-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK26-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK26-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK26-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK26-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i64 10) // CHECK26-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i64 10)
// CHECK26-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0) // CHECK26-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0)
// CHECK26-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK26-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK26-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK26-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK26: omp_offload.failed: // CHECK26: omp_offload.failed:
// CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR5]] // CHECK26-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR5]]
// CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK26-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK26: omp_offload.cont: // CHECK26: omp_offload.cont:
// CHECK26-NEXT: ret i32 0 // CHECK26-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 189 Lines • ▼ Show 20 Lines
// CHECK27-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK27-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK27-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[I_CASTED:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[I_CASTED:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 // CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
// CHECK27-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 4
// CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK27-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK27-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK27-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK27-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK27-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK27-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK27-NEXT: store i32 100, i32* [[N]], align 4 // CHECK27-NEXT: store i32 100, i32* [[N]], align 4
// CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
Show All 10 Lines
// CHECK27-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK27-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK27-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64 // CHECK27-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64
// CHECK27-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK27-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP9]], align 4 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP9]], align 4
// CHECK27-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32* // CHECK27-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32*
// CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP11]], align 4 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP11]], align 4
// CHECK27-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK27-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK27-NEXT: store i64 4, i64* [[TMP12]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP12]], align 4
// CHECK27-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK27-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK27-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32**
// CHECK27-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP14]], align 4
// CHECK27-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32** // CHECK27-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP15]], align 4 // CHECK27-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32**
// CHECK27-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP16]], align 4
// CHECK27-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32** // CHECK27-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 1), align 4
// CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4 // CHECK27-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK27-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK27-NEXT: store i64 [[TMP7]], i64* [[TMP18]], align 4 // CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK27-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK27-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP19]], align 4
// CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK27-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32* // CHECK27-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP21]], align 4 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP21]], align 4
// CHECK27-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK27-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK27-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK27-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP23]], align 4 // CHECK27-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK27-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK27-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK27-NEXT: store i64 4, i64* [[TMP24]], align 4 // CHECK27-NEXT: store i32 [[TMP5]], i32* [[TMP24]], align 4
// CHECK27-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK27-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK27-NEXT: store i8* null, i8** [[TMP25]], align 4 // CHECK27-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK27-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK27-NEXT: store i32 [[TMP5]], i32* [[TMP26]], align 4
// CHECK27-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32* // CHECK27-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK27-NEXT: store i32 [[TMP5]], i32* [[TMP27]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK27-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK27-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK27-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK27-NEXT: store i32 [[TMP5]], i32* [[TMP29]], align 4 // CHECK27-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK27-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK27-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: store i64 4, i64* [[TMP30]], align 4 // CHECK27-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK27-NEXT: store i8* null, i8** [[TMP31]], align 4
// CHECK27-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK27-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK27-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK27-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK27-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK27-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK27-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK27-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK27-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK27-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK27-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK27-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK27-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK27: omp_offload.failed: // CHECK27: omp_offload.failed:
// CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166(i32 [[TMP0]], i32* [[VLA]], i32 [[TMP3]], i32 [[TMP5]]) #[[ATTR5:[0-9]+]] // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166(i32 [[TMP0]], i32* [[VLA]], i32 [[TMP3]], i32 [[TMP5]]) #[[ATTR5:[0-9]+]]
// CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK27: omp_offload.cont: // CHECK27: omp_offload.cont:
// CHECK27-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK27-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK27-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP41]]) // CHECK27-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP36]])
// CHECK27-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK27-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK27-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK27-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK27-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK27-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK27-NEXT: [[TMP43:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK27-NEXT: [[TMP38:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK27-NEXT: ret i32 [[TMP43]] // CHECK27-NEXT: ret i32 [[TMP38]]
// //
// //
// CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166
// CHECK27-SAME: (i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[I:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK27-SAME: (i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[I:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK27-NEXT: entry: // CHECK27-NEXT: entry:
// CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK27-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4 // CHECK27-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 289 Lines • ▼ Show 20 Lines
// CHECK27-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK27-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4
// CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK27-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK27-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK27-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK27-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i64 10) // CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i64 10)
// CHECK27-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0) // CHECK27-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0)
// CHECK27-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK27-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK27-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK27-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK27: omp_offload.failed: // CHECK27: omp_offload.failed:
// CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR5]] // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR5]]
// CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK27: omp_offload.cont: // CHECK27: omp_offload.cont:
// CHECK27-NEXT: ret i32 0 // CHECK27-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 182 Lines • ▼ Show 20 Lines
// CHECK28-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK28-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK28-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[I_CASTED:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[I_CASTED:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK28-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK28-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK28-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK28-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 // CHECK28-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
// CHECK28-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 4
// CHECK28-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK28-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK28-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK28-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK28-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK28-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK28-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK28-NEXT: store i32 100, i32* [[N]], align 4 // CHECK28-NEXT: store i32 100, i32* [[N]], align 4
// CHECK28-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK28-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
Show All 10 Lines
// CHECK28-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK28-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK28-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64 // CHECK28-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64
// CHECK28-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK28-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP9]], align 4 // CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP9]], align 4
// CHECK28-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32* // CHECK28-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32*
// CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP11]], align 4 // CHECK28-NEXT: store i32 [[TMP0]], i32* [[TMP11]], align 4
// CHECK28-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK28-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK28-NEXT: store i64 4, i64* [[TMP12]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP12]], align 4
// CHECK28-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK28-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK28-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32**
// CHECK28-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP14]], align 4
// CHECK28-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32** // CHECK28-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP15]], align 4 // CHECK28-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32**
// CHECK28-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP16]], align 4
// CHECK28-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32** // CHECK28-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 1), align 4
// CHECK28-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4 // CHECK28-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK28-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK28-NEXT: store i64 [[TMP7]], i64* [[TMP18]], align 4 // CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK28-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK28-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK28-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP19]], align 4
// CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK28-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32* // CHECK28-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP21]], align 4 // CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP21]], align 4
// CHECK28-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK28-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK28-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK28-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK28-NEXT: store i32 [[TMP3]], i32* [[TMP23]], align 4 // CHECK28-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK28-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK28-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK28-NEXT: store i64 4, i64* [[TMP24]], align 4 // CHECK28-NEXT: store i32 [[TMP5]], i32* [[TMP24]], align 4
// CHECK28-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK28-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK28-NEXT: store i8* null, i8** [[TMP25]], align 4 // CHECK28-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK28-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK28-NEXT: store i32 [[TMP5]], i32* [[TMP26]], align 4
// CHECK28-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32* // CHECK28-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK28-NEXT: store i32 [[TMP5]], i32* [[TMP27]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK28-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK28-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK28-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK28-NEXT: store i32 [[TMP5]], i32* [[TMP29]], align 4 // CHECK28-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK28-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK28-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: store i64 4, i64* [[TMP30]], align 4 // CHECK28-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK28-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK28-NEXT: store i8* null, i8** [[TMP31]], align 4
// CHECK28-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK28-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK28-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK28-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK28-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK28-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK28-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK28-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK28-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK28-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK28-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK28-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK28-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK28-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK28-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK28-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK28-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK28-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK28-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK28-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK28-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK28-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK28: omp_offload.failed: // CHECK28: omp_offload.failed:
// CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166(i32 [[TMP0]], i32* [[VLA]], i32 [[TMP3]], i32 [[TMP5]]) #[[ATTR5:[0-9]+]] // CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166(i32 [[TMP0]], i32* [[VLA]], i32 [[TMP3]], i32 [[TMP5]]) #[[ATTR5:[0-9]+]]
// CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK28: omp_offload.cont: // CHECK28: omp_offload.cont:
// CHECK28-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK28-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK28-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP41]]) // CHECK28-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP36]])
// CHECK28-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK28-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK28-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK28-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK28-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK28-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK28-NEXT: [[TMP43:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK28-NEXT: [[TMP38:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK28-NEXT: ret i32 [[TMP43]] // CHECK28-NEXT: ret i32 [[TMP38]]
// //
// //
// CHECK28-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166 // CHECK28-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l166
// CHECK28-SAME: (i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[I:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK28-SAME: (i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[I:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK28-NEXT: entry: // CHECK28-NEXT: entry:
// CHECK28-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK28-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK28-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK28-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4 // CHECK28-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 289 Lines • ▼ Show 20 Lines
// CHECK28-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK28-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK28-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 // CHECK28-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4
// CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK28-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK28-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK28-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK28-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK28-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK28-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK28-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i64 10) // CHECK28-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i64 10)
// CHECK28-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0) // CHECK28-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB4]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0)
// CHECK28-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK28-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK28-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK28-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK28: omp_offload.failed: // CHECK28: omp_offload.failed:
// CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR5]] // CHECK28-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l155(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR5]]
// CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK28-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK28: omp_offload.cont: // CHECK28: omp_offload.cont:
// CHECK28-NEXT: ret i32 0 // CHECK28-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 720 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_distribute_parallel_for_simd_collapse_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 197 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: br label [[COND_END]] // CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end: // CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond: // CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body: // CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, %struct.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], %struct.SS* [[TMP0]]) // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, %struct.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], %struct.SS* [[TMP0]]), !llvm.access.group !4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc: // CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK1: omp.inner.for.end: // CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit: // CHECK1: omp.loop.exit:
// CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then: // CHECK1: .omp.final.then:
▲ Show 20 Lines • Show All 50 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: br label [[COND_END]] // CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end: // CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond: // CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !8
// CHECK1-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] // CHECK1-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body: // CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 456 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 456
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !8
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK1-NEXT: [[DIV4:%.*]] = sdiv i32 [[TMP12]], 456 // CHECK1-NEXT: [[DIV4:%.*]] = sdiv i32 [[TMP12]], 456
// CHECK1-NEXT: [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 456 // CHECK1-NEXT: [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 456
// CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL5]] // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL5]]
// CHECK1-NEXT: [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK1-NEXT: [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]] // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]]
// CHECK1-NEXT: store i32 [[ADD7]], i32* [[J]], align 4 // CHECK1-NEXT: store i32 [[ADD7]], i32* [[J]], align 4, !llvm.access.group !8
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !8
// CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i64 0, i64 [[IDXPROM]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !8
// CHECK1-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64 // CHECK1-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64
// CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]] // CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]]
// CHECK1-NEXT: store i32 0, i32* [[ARRAYIDX9]], align 4 // CHECK1-NEXT: store i32 0, i32* [[ARRAYIDX9]], align 4, !llvm.access.group !8
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue: // CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc: // CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK1-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK1-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK1-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1: omp.inner.for.end: // CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit: // CHECK1: omp.loop.exit:
// CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then: // CHECK1: .omp.final.then:
▲ Show 20 Lines • Show All 102 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: br label [[COND_END]] // CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end: // CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond: // CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body: // CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK2-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK2-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, %struct.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], %struct.SS* [[TMP0]]) // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, %struct.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], %struct.SS* [[TMP0]]), !llvm.access.group !4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc: // CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK2: omp.inner.for.end: // CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit: // CHECK2: omp.loop.exit:
// CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK2-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK2-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK2-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK2: .omp.final.then: // CHECK2: .omp.final.then:
▲ Show 20 Lines • Show All 50 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: br label [[COND_END]] // CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end: // CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond: // CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !8
// CHECK2-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] // CHECK2-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK2-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body: // CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 456 // CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 456
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !8
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK2-NEXT: [[DIV4:%.*]] = sdiv i32 [[TMP12]], 456 // CHECK2-NEXT: [[DIV4:%.*]] = sdiv i32 [[TMP12]], 456
// CHECK2-NEXT: [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 456 // CHECK2-NEXT: [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 456
// CHECK2-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL5]] // CHECK2-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL5]]
// CHECK2-NEXT: [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK2-NEXT: [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]] // CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]]
// CHECK2-NEXT: store i32 [[ADD7]], i32* [[J]], align 4 // CHECK2-NEXT: store i32 [[ADD7]], i32* [[J]], align 4, !llvm.access.group !8
// CHECK2-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 // CHECK2-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !8
// CHECK2-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64 // CHECK2-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i64 0, i64 [[IDXPROM]] // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i64 0, i64 [[IDXPROM]]
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4 // CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !8
// CHECK2-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64 // CHECK2-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64
// CHECK2-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]] // CHECK2-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]]
// CHECK2-NEXT: store i32 0, i32* [[ARRAYIDX9]], align 4 // CHECK2-NEXT: store i32 0, i32* [[ARRAYIDX9]], align 4, !llvm.access.group !8
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue: // CHECK2: omp.body.continue:
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc: // CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK2-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK2-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK2-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !8
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK2: omp.inner.for.end: // CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit: // CHECK2: omp.loop.exit:
// CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK2-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK2-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK2-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK2: .omp.final.then: // CHECK2: .omp.final.then:
▲ Show 20 Lines • Show All 102 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: br label [[COND_END]] // CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end: // CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond: // CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body: // CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !5
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, %struct.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], %struct.SS* [[TMP0]]) // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, %struct.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], %struct.SS* [[TMP0]]), !llvm.access.group !5
// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3: omp.inner.for.inc: // CHECK3: omp.inner.for.inc:
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK3: omp.inner.for.end: // CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit: // CHECK3: omp.loop.exit:
// CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK3-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK3-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK3-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK3: .omp.final.then: // CHECK3: .omp.final.then:
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: br label [[COND_END]] // CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end: // CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond: // CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !9
// CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body: // CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 456 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 456
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !9
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK3-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP12]], 456 // CHECK3-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP12]], 456
// CHECK3-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456 // CHECK3-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL4]] // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL4]]
// CHECK3-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK3-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK3-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK3-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !9
// CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !9
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i32 0, i32 [[TMP13]] // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i32 0, i32 [[TMP13]]
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !9
// CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP14]] // CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP14]]
// CHECK3-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4 // CHECK3-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4, !llvm.access.group !9
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3: omp.body.continue: // CHECK3: omp.body.continue:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3: omp.inner.for.inc: // CHECK3: omp.inner.for.inc:
// CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK3-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK3: omp.inner.for.end: // CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit: // CHECK3: omp.loop.exit:
// CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK3-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK3: .omp.final.then: // CHECK3: .omp.final.then:
▲ Show 20 Lines • Show All 102 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: br label [[COND_END]] // CHECK4-NEXT: br label [[COND_END]]
// CHECK4: cond.end: // CHECK4: cond.end:
// CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK4: omp.inner.for.cond: // CHECK4: omp.inner.for.cond:
// CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK4: omp.inner.for.body: // CHECK4: omp.inner.for.body:
// CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !5
// CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, %struct.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], %struct.SS* [[TMP0]]) // CHECK4-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, %struct.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], %struct.SS* [[TMP0]]), !llvm.access.group !5
// CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK4: omp.inner.for.inc: // CHECK4: omp.inner.for.inc:
// CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK4-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK4: omp.inner.for.end: // CHECK4: omp.inner.for.end:
// CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK4: omp.loop.exit: // CHECK4: omp.loop.exit:
// CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK4-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK4-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK4-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK4-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK4: .omp.final.then: // CHECK4: .omp.final.then:
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: br label [[COND_END]] // CHECK4-NEXT: br label [[COND_END]]
// CHECK4: cond.end: // CHECK4: cond.end:
// CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK4-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK4: omp.inner.for.cond: // CHECK4: omp.inner.for.cond:
// CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !9
// CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] // CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK4: omp.inner.for.body: // CHECK4: omp.inner.for.body:
// CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK4-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 456 // CHECK4-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 456
// CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !9
// CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK4-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP12]], 456 // CHECK4-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP12]], 456
// CHECK4-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456 // CHECK4-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK4-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL4]] // CHECK4-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL4]]
// CHECK4-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK4-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK4-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK4-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK4-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK4-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !9
// CHECK4-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 // CHECK4-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
// CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !9
// CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i32 0, i32 [[TMP13]] // CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i32 0, i32 [[TMP13]]
// CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !9
// CHECK4-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP14]] // CHECK4-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP14]]
// CHECK4-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4 // CHECK4-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4, !llvm.access.group !9
// CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK4: omp.body.continue: // CHECK4: omp.body.continue:
// CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK4: omp.inner.for.inc: // CHECK4: omp.inner.for.inc:
// CHECK4-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK4-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK4-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK4-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !9
// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK4: omp.inner.for.end: // CHECK4: omp.inner.for.end:
// CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK4: omp.loop.exit: // CHECK4: omp.loop.exit:
// CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK4-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK4-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK4-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK4-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK4-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK4-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK4: .omp.final.then: // CHECK4: .omp.final.then:
▲ Show 20 Lines • Show All 302 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK9-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK9-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK9-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK9-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK9-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK9-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK9-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK9-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK9-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK9-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK9-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK9-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP54]] // CHECK9-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 94 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: br label [[COND_END]] // CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end: // CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i64 [[COND]], i64* [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: store i64 [[COND]], i64* [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond: // CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK9-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !5
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32*, i64, i64, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP21]], i64 [[TMP22]], i32* [[TMP0]], i32* [[TMP1]], i64 [[TMP2]], i64 [[TMP3]], i32* [[TMP4]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32*, i64, i64, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP21]], i64 [[TMP22]], i32* [[TMP0]], i32* [[TMP1]], i64 [[TMP2]], i64 [[TMP3]], i32* [[TMP4]]), !llvm.access.group !5
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_STRIDE]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_STRIDE]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP23]], [[TMP24]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP23]], [[TMP24]]
// CHECK9-NEXT: store i64 [[ADD]], i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[ADD]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]])
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
▲ Show 20 Lines • Show All 107 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: br label [[COND_END]] // CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end: // CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP18]], [[COND_TRUE]] ], [ [[TMP19]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP18]], [[COND_TRUE]] ], [ [[TMP19]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP20]], i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[TMP20]], i64* [[DOTOMP_IV]], align 8
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond: // CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK9-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !9
// CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP21]], [[TMP22]] // CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP21]], [[TMP22]]
// CHECK9-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !9
// CHECK9-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP24]], 0 // CHECK9-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1 // CHECK9-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK9-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]] // CHECK9-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK9-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64 // CHECK9-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK9-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP23]], [[CONV18]] // CHECK9-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP23]], [[CONV18]]
// CHECK9-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1 // CHECK9-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK9-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32 // CHECK9-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK9-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4 // CHECK9-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4, !llvm.access.group !9
// CHECK9-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK9-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !9
// CHECK9-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP27]], 0 // CHECK9-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP27]], 0
// CHECK9-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1 // CHECK9-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK9-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]] // CHECK9-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK9-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64 // CHECK9-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK9-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP26]], [[CONV25]] // CHECK9-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP26]], [[CONV25]]
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !9
// CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP28]], 0 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]] // CHECK9-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK9-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64 // CHECK9-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]] // CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK9-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP25]], [[MUL31]] // CHECK9-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP25]], [[MUL31]]
// CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1 // CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK9-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]] // CHECK9-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK9-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32 // CHECK9-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK9-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4 // CHECK9-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4, !llvm.access.group !9
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[I11]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[I11]], align 4, !llvm.access.group !9
// CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP29]] to i64 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP29]] to i64
// CHECK9-NEXT: [[TMP30:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]] // CHECK9-NEXT: [[TMP30:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]]
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i64 [[TMP30]] // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i64 [[TMP30]]
// CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[J12]], align 4 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, i32* [[J12]], align 4, !llvm.access.group !9
// CHECK9-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP31]] to i64 // CHECK9-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP31]] to i64
// CHECK9-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 [[IDXPROM36]] // CHECK9-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 [[IDXPROM36]]
// CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX37]], align 4 // CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX37]], align 4, !llvm.access.group !9
// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9: omp.body.continue: // CHECK9: omp.body.continue:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP32:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP32:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK9-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP32]], 1 // CHECK9-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP32]], 1
// CHECK9-NEXT: store i64 [[ADD38]], i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[ADD38]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]])
// CHECK9-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 // CHECK9-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
Show All 35 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: br label [[COND_END]] // CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end: // CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond: // CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !14
// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !14
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !14
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x [2 x i32]]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x [2 x i32]]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x [2 x i32]]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x [2 x i32]]* [[TMP0]]), !llvm.access.group !14
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !14
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
▲ Show 20 Lines • Show All 50 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: br label [[COND_END]] // CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end: // CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond: // CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !17
// CHECK9-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] // CHECK9-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK9-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 2 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 2
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !17
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK9-NEXT: [[DIV4:%.*]] = sdiv i32 [[TMP12]], 2 // CHECK9-NEXT: [[DIV4:%.*]] = sdiv i32 [[TMP12]], 2
// CHECK9-NEXT: [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 2 // CHECK9-NEXT: [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 2
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL5]] // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL5]]
// CHECK9-NEXT: [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK9-NEXT: [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK9-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]] // CHECK9-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]]
// CHECK9-NEXT: store i32 [[ADD7]], i32* [[J]], align 4 // CHECK9-NEXT: store i32 [[ADD7]], i32* [[J]], align 4, !llvm.access.group !17
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !17
// CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i64 0, i64 [[IDXPROM]] // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i64 0, i64 [[IDXPROM]]
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !17
// CHECK9-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64 // CHECK9-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64
// CHECK9-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]] // CHECK9-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]]
// CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX9]], align 4 // CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX9]], align 4, !llvm.access.group !17
// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9: omp.body.continue: // CHECK9: omp.body.continue:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK9-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
Show All 22 Lines
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK10-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK10-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK10-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK10-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK10-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK10-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK10-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK10-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK10-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK10-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK10-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK10-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK10-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK10-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK10-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK10-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK10-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP54]] // CHECK10-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 94 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: br label [[COND_END]] // CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end: // CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ] // CHECK10-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i64 [[COND]], i64* [[DOTOMP_COMB_UB]], align 8 // CHECK10-NEXT: store i64 [[COND]], i64* [[DOTOMP_COMB_UB]], align 8
// CHECK10-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8
// CHECK10-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond: // CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK10-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK10-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8 // CHECK10-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !5
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32*, i64, i64, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP21]], i64 [[TMP22]], i32* [[TMP0]], i32* [[TMP1]], i64 [[TMP2]], i64 [[TMP3]], i32* [[TMP4]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i32*, i64, i64, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP21]], i64 [[TMP22]], i32* [[TMP0]], i32* [[TMP1]], i64 [[TMP2]], i64 [[TMP3]], i32* [[TMP4]]), !llvm.access.group !5
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_STRIDE]], align 8 // CHECK10-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_STRIDE]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP23]], [[TMP24]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP23]], [[TMP24]]
// CHECK10-NEXT: store i64 [[ADD]], i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: store i64 [[ADD]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4 // CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]])
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
▲ Show 20 Lines • Show All 107 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: br label [[COND_END]] // CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end: // CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i64 [ [[TMP18]], [[COND_TRUE]] ], [ [[TMP19]], [[COND_FALSE]] ] // CHECK10-NEXT: [[COND:%.*]] = phi i64 [ [[TMP18]], [[COND_TRUE]] ], [ [[TMP19]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 // CHECK10-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK10-NEXT: store i64 [[TMP20]], i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: store i64 [[TMP20]], i64* [[DOTOMP_IV]], align 8
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond: // CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK10-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !9
// CHECK10-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP21]], [[TMP22]] // CHECK10-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP21]], [[TMP22]]
// CHECK10-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !9
// CHECK10-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP24]], 0 // CHECK10-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1 // CHECK10-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK10-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]] // CHECK10-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK10-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64 // CHECK10-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK10-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP23]], [[CONV18]] // CHECK10-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP23]], [[CONV18]]
// CHECK10-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1 // CHECK10-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK10-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32 // CHECK10-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK10-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4 // CHECK10-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4, !llvm.access.group !9
// CHECK10-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK10-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !9
// CHECK10-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP27]], 0 // CHECK10-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP27]], 0
// CHECK10-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1 // CHECK10-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK10-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]] // CHECK10-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK10-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64 // CHECK10-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK10-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP26]], [[CONV25]] // CHECK10-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP26]], [[CONV25]]
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !9
// CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP28]], 0 // CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1 // CHECK10-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK10-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]] // CHECK10-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK10-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64 // CHECK10-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK10-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]] // CHECK10-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK10-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP25]], [[MUL31]] // CHECK10-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP25]], [[MUL31]]
// CHECK10-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1 // CHECK10-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK10-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]] // CHECK10-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK10-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32 // CHECK10-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK10-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4 // CHECK10-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4, !llvm.access.group !9
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[I11]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[I11]], align 4, !llvm.access.group !9
// CHECK10-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP29]] to i64 // CHECK10-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP29]] to i64
// CHECK10-NEXT: [[TMP30:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]] // CHECK10-NEXT: [[TMP30:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]]
// CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i64 [[TMP30]] // CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i64 [[TMP30]]
// CHECK10-NEXT: [[TMP31:%.*]] = load i32, i32* [[J12]], align 4 // CHECK10-NEXT: [[TMP31:%.*]] = load i32, i32* [[J12]], align 4, !llvm.access.group !9
// CHECK10-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP31]] to i64 // CHECK10-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP31]] to i64
// CHECK10-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 [[IDXPROM36]] // CHECK10-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 [[IDXPROM36]]
// CHECK10-NEXT: store i32 0, i32* [[ARRAYIDX37]], align 4 // CHECK10-NEXT: store i32 0, i32* [[ARRAYIDX37]], align 4, !llvm.access.group !9
// CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10: omp.body.continue: // CHECK10: omp.body.continue:
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP32:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP32:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK10-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP32]], 1 // CHECK10-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP32]], 1
// CHECK10-NEXT: store i64 [[ADD38]], i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: store i64 [[ADD38]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !9
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 // CHECK10-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]])
// CHECK10-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 // CHECK10-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
Show All 35 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: br label [[COND_END]] // CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end: // CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond: // CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !14
// CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !14
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !14
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x [2 x i32]]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x [2 x i32]]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x [2 x i32]]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x [2 x i32]]* [[TMP0]]), !llvm.access.group !14
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !14
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !14
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK10: .omp.final.then: // CHECK10: .omp.final.then:
▲ Show 20 Lines • Show All 50 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: br label [[COND_END]] // CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end: // CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond: // CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !17
// CHECK10-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] // CHECK10-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK10-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 2 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 2
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !17
// CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK10-NEXT: [[DIV4:%.*]] = sdiv i32 [[TMP12]], 2 // CHECK10-NEXT: [[DIV4:%.*]] = sdiv i32 [[TMP12]], 2
// CHECK10-NEXT: [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 2 // CHECK10-NEXT: [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 2
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL5]] // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL5]]
// CHECK10-NEXT: [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK10-NEXT: [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK10-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]] // CHECK10-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]]
// CHECK10-NEXT: store i32 [[ADD7]], i32* [[J]], align 4 // CHECK10-NEXT: store i32 [[ADD7]], i32* [[J]], align 4, !llvm.access.group !17
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !17
// CHECK10-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64 // CHECK10-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i64 0, i64 [[IDXPROM]] // CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i64 0, i64 [[IDXPROM]]
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !17
// CHECK10-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64 // CHECK10-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64
// CHECK10-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]] // CHECK10-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]]
// CHECK10-NEXT: store i32 0, i32* [[ARRAYIDX9]], align 4 // CHECK10-NEXT: store i32 0, i32* [[ARRAYIDX9]], align 4, !llvm.access.group !17
// CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10: omp.body.continue: // CHECK10: omp.body.continue:
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK10-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK10-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK10-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !17
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK10-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK10-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK10-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK10: .omp.final.then: // CHECK10: .omp.final.then:
Show All 22 Lines
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK11-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK11-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK11-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK11-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK11-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK11-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK11-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK11-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK11-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK11-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP53]] // CHECK11-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 92 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: br label [[COND_END]] // CHECK11-NEXT: br label [[COND_END]]
// CHECK11: cond.end: // CHECK11: cond.end:
// CHECK11-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ] // CHECK11-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i64 [[COND]], i64* [[DOTOMP_COMB_UB]], align 8 // CHECK11-NEXT: store i64 [[COND]], i64* [[DOTOMP_COMB_UB]], align 8
// CHECK11-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8 // CHECK11-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8
// CHECK11-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8 // CHECK11-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK11-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK11-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8 // CHECK11-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[TMP22:%.*]] = trunc i64 [[TMP21]] to i32 // CHECK11-NEXT: [[TMP22:%.*]] = trunc i64 [[TMP21]] to i32
// CHECK11-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8 // CHECK11-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[TMP24:%.*]] = trunc i64 [[TMP23]] to i32 // CHECK11-NEXT: [[TMP24:%.*]] = trunc i64 [[TMP23]] to i32
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32*, i32, i32, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP22]], i32 [[TMP24]], i32* [[TMP0]], i32* [[TMP1]], i32 [[TMP2]], i32 [[TMP3]], i32* [[TMP4]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32*, i32, i32, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP22]], i32 [[TMP24]], i32* [[TMP0]], i32* [[TMP1]], i32 [[TMP2]], i32 [[TMP3]], i32* [[TMP4]]), !llvm.access.group !6
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_STRIDE]], align 8 // CHECK11-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_STRIDE]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP25]], [[TMP26]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP25]], [[TMP26]]
// CHECK11-NEXT: store i64 [[ADD]], i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: store i64 [[ADD]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK11-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP28]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP28]])
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK11-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
▲ Show 20 Lines • Show All 109 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: br label [[COND_END]] // CHECK11-NEXT: br label [[COND_END]]
// CHECK11: cond.end: // CHECK11: cond.end:
// CHECK11-NEXT: [[COND:%.*]] = phi i64 [ [[TMP18]], [[COND_TRUE]] ], [ [[TMP19]], [[COND_FALSE]] ] // CHECK11-NEXT: [[COND:%.*]] = phi i64 [ [[TMP18]], [[COND_TRUE]] ], [ [[TMP19]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 // CHECK11-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK11-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 // CHECK11-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK11-NEXT: store i64 [[TMP20]], i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: store i64 [[TMP20]], i64* [[DOTOMP_IV]], align 8
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK11-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 // CHECK11-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !10
// CHECK11-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP21]], [[TMP22]] // CHECK11-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP21]], [[TMP22]]
// CHECK11-NEXT: br i1 [[CMP16]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP16]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !10
// CHECK11-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP24]], 0 // CHECK11-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[DIV18:%.*]] = sdiv i32 [[SUB17]], 1 // CHECK11-NEXT: [[DIV18:%.*]] = sdiv i32 [[SUB17]], 1
// CHECK11-NEXT: [[MUL19:%.*]] = mul nsw i32 1, [[DIV18]] // CHECK11-NEXT: [[MUL19:%.*]] = mul nsw i32 1, [[DIV18]]
// CHECK11-NEXT: [[CONV20:%.*]] = sext i32 [[MUL19]] to i64 // CHECK11-NEXT: [[CONV20:%.*]] = sext i32 [[MUL19]] to i64
// CHECK11-NEXT: [[DIV21:%.*]] = sdiv i64 [[TMP23]], [[CONV20]] // CHECK11-NEXT: [[DIV21:%.*]] = sdiv i64 [[TMP23]], [[CONV20]]
// CHECK11-NEXT: [[MUL22:%.*]] = mul nsw i64 [[DIV21]], 1 // CHECK11-NEXT: [[MUL22:%.*]] = mul nsw i64 [[DIV21]], 1
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL22]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL22]]
// CHECK11-NEXT: [[CONV23:%.*]] = trunc i64 [[ADD]] to i32 // CHECK11-NEXT: [[CONV23:%.*]] = trunc i64 [[ADD]] to i32
// CHECK11-NEXT: store i32 [[CONV23]], i32* [[I13]], align 4 // CHECK11-NEXT: store i32 [[CONV23]], i32* [[I13]], align 4, !llvm.access.group !10
// CHECK11-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK11-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !10
// CHECK11-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP27]], 0 // CHECK11-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP27]], 0
// CHECK11-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 // CHECK11-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK11-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]] // CHECK11-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]]
// CHECK11-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64 // CHECK11-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64
// CHECK11-NEXT: [[DIV28:%.*]] = sdiv i64 [[TMP26]], [[CONV27]] // CHECK11-NEXT: [[DIV28:%.*]] = sdiv i64 [[TMP26]], [[CONV27]]
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !10
// CHECK11-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP28]], 0 // CHECK11-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1 // CHECK11-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK11-NEXT: [[MUL31:%.*]] = mul nsw i32 1, [[DIV30]] // CHECK11-NEXT: [[MUL31:%.*]] = mul nsw i32 1, [[DIV30]]
// CHECK11-NEXT: [[CONV32:%.*]] = sext i32 [[MUL31]] to i64 // CHECK11-NEXT: [[CONV32:%.*]] = sext i32 [[MUL31]] to i64
// CHECK11-NEXT: [[MUL33:%.*]] = mul nsw i64 [[DIV28]], [[CONV32]] // CHECK11-NEXT: [[MUL33:%.*]] = mul nsw i64 [[DIV28]], [[CONV32]]
// CHECK11-NEXT: [[SUB34:%.*]] = sub nsw i64 [[TMP25]], [[MUL33]] // CHECK11-NEXT: [[SUB34:%.*]] = sub nsw i64 [[TMP25]], [[MUL33]]
// CHECK11-NEXT: [[MUL35:%.*]] = mul nsw i64 [[SUB34]], 1 // CHECK11-NEXT: [[MUL35:%.*]] = mul nsw i64 [[SUB34]], 1
// CHECK11-NEXT: [[ADD36:%.*]] = add nsw i64 0, [[MUL35]] // CHECK11-NEXT: [[ADD36:%.*]] = add nsw i64 0, [[MUL35]]
// CHECK11-NEXT: [[CONV37:%.*]] = trunc i64 [[ADD36]] to i32 // CHECK11-NEXT: [[CONV37:%.*]] = trunc i64 [[ADD36]] to i32
// CHECK11-NEXT: store i32 [[CONV37]], i32* [[J14]], align 4 // CHECK11-NEXT: store i32 [[CONV37]], i32* [[J14]], align 4, !llvm.access.group !10
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[I13]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[I13]], align 4, !llvm.access.group !10
// CHECK11-NEXT: [[TMP30:%.*]] = mul nsw i32 [[TMP29]], [[TMP3]] // CHECK11-NEXT: [[TMP30:%.*]] = mul nsw i32 [[TMP29]], [[TMP3]]
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i32 [[TMP30]] // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i32 [[TMP30]]
// CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[J14]], align 4 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, i32* [[J14]], align 4, !llvm.access.group !10
// CHECK11-NEXT: [[ARRAYIDX38:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i32 [[TMP31]] // CHECK11-NEXT: [[ARRAYIDX38:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i32 [[TMP31]]
// CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX38]], align 4 // CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX38]], align 4, !llvm.access.group !10
// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11: omp.body.continue: // CHECK11: omp.body.continue:
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP32:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP32:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK11-NEXT: [[ADD39:%.*]] = add nsw i64 [[TMP32]], 1 // CHECK11-NEXT: [[ADD39:%.*]] = add nsw i64 [[TMP32]], 1
// CHECK11-NEXT: store i64 [[ADD39]], i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: store i64 [[ADD39]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK11-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 // CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]])
// CHECK11-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 // CHECK11-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
Show All 35 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: br label [[COND_END]] // CHECK11-NEXT: br label [[COND_END]]
// CHECK11: cond.end: // CHECK11: cond.end:
// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !15
// CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !15
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !15
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x [2 x i32]]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x [2 x i32]]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x [2 x i32]]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x [2 x i32]]* [[TMP0]]), !llvm.access.group !15
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !15
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK11-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK11-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK11-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then: // CHECK11: .omp.final.then:
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: br label [[COND_END]] // CHECK11-NEXT: br label [[COND_END]]
// CHECK11: cond.end: // CHECK11: cond.end:
// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !18
// CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 2 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 2
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !18
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK11-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP12]], 2 // CHECK11-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP12]], 2
// CHECK11-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2 // CHECK11-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL4]] // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL4]]
// CHECK11-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK11-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK11-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK11-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !18
// CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !18
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i32 0, i32 [[TMP13]] // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i32 0, i32 [[TMP13]]
// CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !18
// CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP14]] // CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP14]]
// CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4 // CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4, !llvm.access.group !18
// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11: omp.body.continue: // CHECK11: omp.body.continue:
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK11-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK11-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then: // CHECK11: .omp.final.then:
Show All 22 Lines
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK12-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK12-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK12-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK12-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK12-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK12-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK12-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK12-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK12-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK12-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK12-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK12-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK12-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK12-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK12-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK12-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK12-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP53]] // CHECK12-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l86
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 92 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: br label [[COND_END]] // CHECK12-NEXT: br label [[COND_END]]
// CHECK12: cond.end: // CHECK12: cond.end:
// CHECK12-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ] // CHECK12-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK12-NEXT: store i64 [[COND]], i64* [[DOTOMP_COMB_UB]], align 8 // CHECK12-NEXT: store i64 [[COND]], i64* [[DOTOMP_COMB_UB]], align 8
// CHECK12-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8 // CHECK12-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8
// CHECK12-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8 // CHECK12-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK12-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK12-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8 // CHECK12-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_COMB_LB]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[TMP22:%.*]] = trunc i64 [[TMP21]] to i32 // CHECK12-NEXT: [[TMP22:%.*]] = trunc i64 [[TMP21]] to i32
// CHECK12-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8 // CHECK12-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[TMP24:%.*]] = trunc i64 [[TMP23]] to i32 // CHECK12-NEXT: [[TMP24:%.*]] = trunc i64 [[TMP23]] to i32
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32*, i32, i32, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP22]], i32 [[TMP24]], i32* [[TMP0]], i32* [[TMP1]], i32 [[TMP2]], i32 [[TMP3]], i32* [[TMP4]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32*, i32, i32, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP22]], i32 [[TMP24]], i32* [[TMP0]], i32* [[TMP1]], i32 [[TMP2]], i32 [[TMP3]], i32* [[TMP4]]), !llvm.access.group !6
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_STRIDE]], align 8 // CHECK12-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_STRIDE]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP25]], [[TMP26]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP25]], [[TMP26]]
// CHECK12-NEXT: store i64 [[ADD]], i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: store i64 [[ADD]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK12-NEXT: [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 // CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP28]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP28]])
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK12-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
▲ Show 20 Lines • Show All 109 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: br label [[COND_END]] // CHECK12-NEXT: br label [[COND_END]]
// CHECK12: cond.end: // CHECK12: cond.end:
// CHECK12-NEXT: [[COND:%.*]] = phi i64 [ [[TMP18]], [[COND_TRUE]] ], [ [[TMP19]], [[COND_FALSE]] ] // CHECK12-NEXT: [[COND:%.*]] = phi i64 [ [[TMP18]], [[COND_TRUE]] ], [ [[TMP19]], [[COND_FALSE]] ]
// CHECK12-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 // CHECK12-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK12-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 // CHECK12-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK12-NEXT: store i64 [[TMP20]], i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: store i64 [[TMP20]], i64* [[DOTOMP_IV]], align 8
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK12-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 // CHECK12-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !10
// CHECK12-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP21]], [[TMP22]] // CHECK12-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP21]], [[TMP22]]
// CHECK12-NEXT: br i1 [[CMP16]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP16]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !10
// CHECK12-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP24]], 0 // CHECK12-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[DIV18:%.*]] = sdiv i32 [[SUB17]], 1 // CHECK12-NEXT: [[DIV18:%.*]] = sdiv i32 [[SUB17]], 1
// CHECK12-NEXT: [[MUL19:%.*]] = mul nsw i32 1, [[DIV18]] // CHECK12-NEXT: [[MUL19:%.*]] = mul nsw i32 1, [[DIV18]]
// CHECK12-NEXT: [[CONV20:%.*]] = sext i32 [[MUL19]] to i64 // CHECK12-NEXT: [[CONV20:%.*]] = sext i32 [[MUL19]] to i64
// CHECK12-NEXT: [[DIV21:%.*]] = sdiv i64 [[TMP23]], [[CONV20]] // CHECK12-NEXT: [[DIV21:%.*]] = sdiv i64 [[TMP23]], [[CONV20]]
// CHECK12-NEXT: [[MUL22:%.*]] = mul nsw i64 [[DIV21]], 1 // CHECK12-NEXT: [[MUL22:%.*]] = mul nsw i64 [[DIV21]], 1
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL22]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL22]]
// CHECK12-NEXT: [[CONV23:%.*]] = trunc i64 [[ADD]] to i32 // CHECK12-NEXT: [[CONV23:%.*]] = trunc i64 [[ADD]] to i32
// CHECK12-NEXT: store i32 [[CONV23]], i32* [[I13]], align 4 // CHECK12-NEXT: store i32 [[CONV23]], i32* [[I13]], align 4, !llvm.access.group !10
// CHECK12-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP25:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK12-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP26:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !10
// CHECK12-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP27]], 0 // CHECK12-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP27]], 0
// CHECK12-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 // CHECK12-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK12-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]] // CHECK12-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]]
// CHECK12-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64 // CHECK12-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64
// CHECK12-NEXT: [[DIV28:%.*]] = sdiv i64 [[TMP26]], [[CONV27]] // CHECK12-NEXT: [[DIV28:%.*]] = sdiv i64 [[TMP26]], [[CONV27]]
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !10
// CHECK12-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP28]], 0 // CHECK12-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1 // CHECK12-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK12-NEXT: [[MUL31:%.*]] = mul nsw i32 1, [[DIV30]] // CHECK12-NEXT: [[MUL31:%.*]] = mul nsw i32 1, [[DIV30]]
// CHECK12-NEXT: [[CONV32:%.*]] = sext i32 [[MUL31]] to i64 // CHECK12-NEXT: [[CONV32:%.*]] = sext i32 [[MUL31]] to i64
// CHECK12-NEXT: [[MUL33:%.*]] = mul nsw i64 [[DIV28]], [[CONV32]] // CHECK12-NEXT: [[MUL33:%.*]] = mul nsw i64 [[DIV28]], [[CONV32]]
// CHECK12-NEXT: [[SUB34:%.*]] = sub nsw i64 [[TMP25]], [[MUL33]] // CHECK12-NEXT: [[SUB34:%.*]] = sub nsw i64 [[TMP25]], [[MUL33]]
// CHECK12-NEXT: [[MUL35:%.*]] = mul nsw i64 [[SUB34]], 1 // CHECK12-NEXT: [[MUL35:%.*]] = mul nsw i64 [[SUB34]], 1
// CHECK12-NEXT: [[ADD36:%.*]] = add nsw i64 0, [[MUL35]] // CHECK12-NEXT: [[ADD36:%.*]] = add nsw i64 0, [[MUL35]]
// CHECK12-NEXT: [[CONV37:%.*]] = trunc i64 [[ADD36]] to i32 // CHECK12-NEXT: [[CONV37:%.*]] = trunc i64 [[ADD36]] to i32
// CHECK12-NEXT: store i32 [[CONV37]], i32* [[J14]], align 4 // CHECK12-NEXT: store i32 [[CONV37]], i32* [[J14]], align 4, !llvm.access.group !10
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[I13]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[I13]], align 4, !llvm.access.group !10
// CHECK12-NEXT: [[TMP30:%.*]] = mul nsw i32 [[TMP29]], [[TMP3]] // CHECK12-NEXT: [[TMP30:%.*]] = mul nsw i32 [[TMP29]], [[TMP3]]
// CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i32 [[TMP30]] // CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i32 [[TMP30]]
// CHECK12-NEXT: [[TMP31:%.*]] = load i32, i32* [[J14]], align 4 // CHECK12-NEXT: [[TMP31:%.*]] = load i32, i32* [[J14]], align 4, !llvm.access.group !10
// CHECK12-NEXT: [[ARRAYIDX38:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i32 [[TMP31]] // CHECK12-NEXT: [[ARRAYIDX38:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i32 [[TMP31]]
// CHECK12-NEXT: store i32 0, i32* [[ARRAYIDX38]], align 4 // CHECK12-NEXT: store i32 0, i32* [[ARRAYIDX38]], align 4, !llvm.access.group !10
// CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK12: omp.body.continue: // CHECK12: omp.body.continue:
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP32:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP32:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK12-NEXT: [[ADD39:%.*]] = add nsw i64 [[TMP32]], 1 // CHECK12-NEXT: [[ADD39:%.*]] = add nsw i64 [[TMP32]], 1
// CHECK12-NEXT: store i64 [[ADD39]], i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: store i64 [[ADD39]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !10
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK12-NEXT: [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4 // CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP34]])
// CHECK12-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK12-NEXT: [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK12-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 // CHECK12-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0
Show All 35 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 20)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l72([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: br label [[COND_END]] // CHECK12-NEXT: br label [[COND_END]]
// CHECK12: cond.end: // CHECK12: cond.end:
// CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !15
// CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !15
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !15
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x [2 x i32]]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x [2 x i32]]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x [2 x i32]]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x [2 x i32]]* [[TMP0]]), !llvm.access.group !15
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !15
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !15
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK12-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK12-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK12-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK12: .omp.final.then: // CHECK12: .omp.final.then:
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: br label [[COND_END]] // CHECK12-NEXT: br label [[COND_END]]
// CHECK12: cond.end: // CHECK12: cond.end:
// CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !18
// CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] // CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 2 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP10]], 2
// CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !18
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK12-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP12]], 2 // CHECK12-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP12]], 2
// CHECK12-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2 // CHECK12-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL4]] // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL4]]
// CHECK12-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK12-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK12-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK12-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK12-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK12-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !18
// CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !18
// CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i32 0, i32 [[TMP13]] // CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i32 0, i32 [[TMP13]]
// CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !18
// CHECK12-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP14]] // CHECK12-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP14]]
// CHECK12-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4 // CHECK12-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4, !llvm.access.group !18
// CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK12: omp.body.continue: // CHECK12: omp.body.continue:
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK12-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK12-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK12-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !18
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK12-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK12-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK12-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK12-NEXT: br i1 [[TMP17]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK12: .omp.final.then: // CHECK12: .omp.final.then:
▲ Show 20 Lines • Show All 813 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_distribute_parallel_for_simd_dist_schedule_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 3,062 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [4 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES24:%.*]] = alloca [4 x i64], align 8 // CHECK9-NEXT: [[_TMP23:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP25:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: store i32 10, i32* [[M]], align 4 // CHECK9-NEXT: store i32 10, i32* [[M]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK9-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK9-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK9-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK9-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK9-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 8
// CHECK9-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK9-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK9-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK9-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK9-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK9-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK9-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK9-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP52]])
// CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK9-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK9: omp_offload.failed15:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK9: omp_offload.cont16:
// CHECK9-NEXT: [[TMP55:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: [[CONV17:%.*]] = bitcast i64* [[M_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP55]], i32* [[CONV17]], align 4
// CHECK9-NEXT: [[TMP56:%.*]] = load i64, i64* [[M_CASTED]], align 8
// CHECK9-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9-NEXT: store i32 [[TMP57]], i32* [[CONV19]], align 4
// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK9-NEXT: [[TMP58:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK9-NEXT: [[TMP59:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK9-NEXT: [[TMP60:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK9-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i64*
// CHECK9-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: store i64 [[TMP56]], i64* [[TMP61]], align 8
// CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK9-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK9-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i64*
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP60]]) // CHECK9-NEXT: store i64 [[TMP56]], i64* [[TMP63]], align 8
// CHECK9-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK9-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK9-NEXT: store i8* null, i8** [[TMP64]], align 8
// CHECK9-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK9-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK9: omp_offload.failed16: // CHECK9-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i64*
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK9-NEXT: store i64 [[TMP58]], i64* [[TMP66]], align 8
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK9-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK9: omp_offload.cont17: // CHECK9-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i64*
// CHECK9-NEXT: [[TMP63:%.*]] = load i32, i32* [[M]], align 4 // CHECK9-NEXT: store i64 [[TMP58]], i64* [[TMP68]], align 8
// CHECK9-NEXT: [[CONV18:%.*]] = bitcast i64* [[M_CASTED]] to i32* // CHECK9-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK9-NEXT: store i32 [[TMP63]], i32* [[CONV18]], align 4 // CHECK9-NEXT: store i8* null, i8** [[TMP69]], align 8
// CHECK9-NEXT: [[TMP64:%.*]] = load i64, i64* [[M_CASTED]], align 8 // CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK9-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32*
// CHECK9-NEXT: store i32 [[TMP65]], i32* [[CONV20]], align 4
// CHECK9-NEXT: [[TMP66:%.*]] = load i64, i64* [[N_CASTED19]], align 8
// CHECK9-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK9-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK9-NEXT: store i64 [[TMP64]], i64* [[TMP69]], align 8
// CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK9-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64* // CHECK9-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64*
// CHECK9-NEXT: store i64 [[TMP64]], i64* [[TMP71]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP71]], align 8
// CHECK9-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 0 // CHECK9-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK9-NEXT: store i64 4, i64* [[TMP72]], align 8 // CHECK9-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i64*
// CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP73]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP73]], align 8 // CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1 // CHECK9-NEXT: store i8* null, i8** [[TMP74]], align 8
// CHECK9-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64* // CHECK9-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP66]], i64* [[TMP75]], align 8 // CHECK9-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 8
// CHECK9-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64* // CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP66]], i64* [[TMP77]], align 8 // CHECK9-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32**
// CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP78]], align 8
// CHECK9-NEXT: store i64 4, i64* [[TMP78]], align 8 // CHECK9-NEXT: store i64 [[TMP59]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1 // CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK9-NEXT: store i8* null, i8** [[TMP79]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP79]], align 8
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2 // CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK9-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64* // CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP81]], align 8 // CHECK9-NEXT: [[TMP82:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2 // CHECK9-NEXT: store i32 [[TMP82]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK9-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i64* // CHECK9-NEXT: [[TMP83:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP83]], align 8 // CHECK9-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP83]], 0
// CHECK9-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 2 // CHECK9-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK9-NEXT: store i64 8, i64* [[TMP84]], align 8 // CHECK9-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK9-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2 // CHECK9-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP85]], align 8 // CHECK9-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3 // CHECK9-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP84]], 1
// CHECK9-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to i32** // CHECK9-NEXT: [[TMP85:%.*]] = zext i32 [[ADD29]] to i64
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP87]], align 8 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP85]])
// CHECK9-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3 // CHECK9-NEXT: [[TMP86:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118.region_id, i32 4, i8** [[TMP80]], i8** [[TMP81]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to i32** // CHECK9-NEXT: [[TMP87:%.*]] = icmp ne i32 [[TMP86]], 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP89]], align 8 // CHECK9-NEXT: br i1 [[TMP87]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK9-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 3 // CHECK9: omp_offload.failed30:
// CHECK9-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118(i64 [[TMP56]], i64 [[TMP58]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK9-NEXT: store i8* null, i8** [[TMP91]], align 8 // CHECK9: omp_offload.cont31:
// CHECK9-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 // CHECK9-NEXT: [[TMP88:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP88]])
// CHECK9-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 0
// CHECK9-NEXT: [[TMP95:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP95]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[TMP96:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[SUB28:%.*]] = sub nsw i32 [[TMP96]], 0
// CHECK9-NEXT: [[DIV29:%.*]] = sdiv i32 [[SUB28]], 1
// CHECK9-NEXT: [[SUB30:%.*]] = sub nsw i32 [[DIV29]], 1
// CHECK9-NEXT: store i32 [[SUB30]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK9-NEXT: [[ADD31:%.*]] = add nsw i32 [[TMP97]], 1
// CHECK9-NEXT: [[TMP98:%.*]] = zext i32 [[ADD31]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP98]])
// CHECK9-NEXT: [[TMP99:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118.region_id, i32 4, i8** [[TMP92]], i8** [[TMP93]], i64* [[TMP94]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP100:%.*]] = icmp ne i32 [[TMP99]], 0
// CHECK9-NEXT: br i1 [[TMP100]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]]
// CHECK9: omp_offload.failed32:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118(i64 [[TMP64]], i64 [[TMP66]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT33]]
// CHECK9: omp_offload.cont33:
// CHECK9-NEXT: [[TMP101:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP101]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP102:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP89:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP102]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP89]])
// CHECK9-NEXT: [[TMP103:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP90:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP103]] // CHECK9-NEXT: ret i32 [[TMP90]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 481 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV2]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV2]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* [[CONV1]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[CONV1]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 65 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
// CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 // CHECK9-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
// CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[CONV7:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV7:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP21]], i32* [[CONV7]], align 4, !llvm.access.group !24 // CHECK9-NEXT: store i32 [[TMP21]], i32* [[CONV7]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !24 // CHECK9-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !24
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i64 [[TMP1]], i32* [[TMP2]], i64 [[TMP22]]), !llvm.access.group !24 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64, i32*, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i64 [[TMP1]], i32* [[TMP2]], i64 [[TMP22]]), !llvm.access.group !24
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
// CHECK9-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK9-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24
// CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24
Show All 37 Lines
// CHECK9-NEXT: store i32 [[ADD18]], i32* [[I4]], align 4 // CHECK9-NEXT: store i32 [[ADD18]], i32* [[I4]], align 4
// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK9: .omp.final.done: // CHECK9: .omp.final.done:
// CHECK9-NEXT: br label [[OMP_PRECOND_END]] // CHECK9-NEXT: br label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.end: // CHECK9: omp.precond.end:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 135 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK9: omp_offload.failed5: // CHECK9: omp_offload.failed5:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK9: omp_offload.cont6: // CHECK9: omp_offload.cont6:
// CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[M_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[M_CASTED]] to i32*
Show All 13 Lines
// CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]** // CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 8
// CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK9-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 // CHECK9-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// CHECK9-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK9-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK9: omp_offload.failed11: // CHECK9: omp_offload.failed11:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96(i64 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96(i64 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK9: omp_offload.cont12: // CHECK9: omp_offload.cont12:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !30 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !30
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !30 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !30
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !30 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !30
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
// CHECK9-NEXT: store i32 10, i32* [[I]], align 4 // CHECK9-NEXT: store i32 10, i32* [[I]], align 4
// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK9: .omp.final.done: // CHECK9: .omp.final.done:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !36 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !36
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !36 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !36
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !36 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !36
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
// CHECK9-NEXT: store i32 10, i32* [[I]], align 4 // CHECK9-NEXT: store i32 10, i32* [[I]], align 4
// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK9: .omp.final.done: // CHECK9: .omp.final.done:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[M_ADDR]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[M_ADDR]] to i32*
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4 // CHECK9-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP3]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP3]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 39 Lines
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4, !llvm.access.group !42 // CHECK9-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !42 // CHECK9-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !42
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]), !llvm.access.group !42 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]), !llvm.access.group !42
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42
// CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42
Show All 26 Lines
// CHECK9-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
// CHECK9-NEXT: store i32 10, i32* [[I]], align 4 // CHECK9-NEXT: store i32 10, i32* [[I]], align 4
// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK9: .omp.final.done: // CHECK9: .omp.final.done:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 91 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK10-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS21:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS22:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS23:%.*]] = alloca [4 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [4 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES24:%.*]] = alloca [4 x i64], align 8 // CHECK10-NEXT: [[_TMP23:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP25:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_27:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: store i32 10, i32* [[M]], align 4 // CHECK10-NEXT: store i32 10, i32* [[M]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK10-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK10-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK10-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK10-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK10-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK10-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 8
// CHECK10-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK10-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK10-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK10-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK10-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK10-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK10-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK10-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK10-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK10-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK10-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK10-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK10-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK10-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK10-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP52]])
// CHECK10-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK10-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK10: omp_offload.failed15:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK10: omp_offload.cont16:
// CHECK10-NEXT: [[TMP55:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: [[CONV17:%.*]] = bitcast i64* [[M_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP55]], i32* [[CONV17]], align 4
// CHECK10-NEXT: [[TMP56:%.*]] = load i64, i64* [[M_CASTED]], align 8
// CHECK10-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32*
// CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10-NEXT: store i32 [[TMP57]], i32* [[CONV19]], align 4
// CHECK10-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK10-NEXT: [[TMP58:%.*]] = load i64, i64* [[N_CASTED18]], align 8
// CHECK10-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK10-NEXT: [[TMP59:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK10-NEXT: [[TMP60:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK10-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i64*
// CHECK10-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: store i64 [[TMP56]], i64* [[TMP61]], align 8
// CHECK10-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK10-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK10-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i64*
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP60]]) // CHECK10-NEXT: store i64 [[TMP56]], i64* [[TMP63]], align 8
// CHECK10-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK10-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK10-NEXT: store i8* null, i8** [[TMP64]], align 8
// CHECK10-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK10-NEXT: [[TMP65:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1
// CHECK10: omp_offload.failed16: // CHECK10-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i64*
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK10-NEXT: store i64 [[TMP58]], i64* [[TMP66]], align 8
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK10-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1
// CHECK10: omp_offload.cont17: // CHECK10-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i64*
// CHECK10-NEXT: [[TMP63:%.*]] = load i32, i32* [[M]], align 4 // CHECK10-NEXT: store i64 [[TMP58]], i64* [[TMP68]], align 8
// CHECK10-NEXT: [[CONV18:%.*]] = bitcast i64* [[M_CASTED]] to i32* // CHECK10-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1
// CHECK10-NEXT: store i32 [[TMP63]], i32* [[CONV18]], align 4 // CHECK10-NEXT: store i8* null, i8** [[TMP69]], align 8
// CHECK10-NEXT: [[TMP64:%.*]] = load i64, i64* [[M_CASTED]], align 8 // CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2
// CHECK10-NEXT: [[TMP65:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV20:%.*]] = bitcast i64* [[N_CASTED19]] to i32*
// CHECK10-NEXT: store i32 [[TMP65]], i32* [[CONV20]], align 4
// CHECK10-NEXT: [[TMP66:%.*]] = load i64, i64* [[N_CASTED19]], align 8
// CHECK10-NEXT: [[TMP67:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP68:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0
// CHECK10-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64*
// CHECK10-NEXT: store i64 [[TMP64]], i64* [[TMP69]], align 8
// CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0
// CHECK10-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64* // CHECK10-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i64*
// CHECK10-NEXT: store i64 [[TMP64]], i64* [[TMP71]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP71]], align 8
// CHECK10-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 0 // CHECK10-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2
// CHECK10-NEXT: store i64 4, i64* [[TMP72]], align 8 // CHECK10-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i64*
// CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 0 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP73]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP73]], align 8 // CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2
// CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 1 // CHECK10-NEXT: store i8* null, i8** [[TMP74]], align 8
// CHECK10-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64* // CHECK10-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3
// CHECK10-NEXT: store i64 [[TMP66]], i64* [[TMP75]], align 8 // CHECK10-NEXT: [[TMP76:%.*]] = bitcast i8** [[TMP75]] to i32**
// CHECK10-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP76]], align 8
// CHECK10-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64* // CHECK10-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 3
// CHECK10-NEXT: store i64 [[TMP66]], i64* [[TMP77]], align 8 // CHECK10-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32**
// CHECK10-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP78]], align 8
// CHECK10-NEXT: store i64 4, i64* [[TMP78]], align 8 // CHECK10-NEXT: store i64 [[TMP59]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 3), align 8
// CHECK10-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 1 // CHECK10-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3
// CHECK10-NEXT: store i8* null, i8** [[TMP79]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP79]], align 8
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 2 // CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK10-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i64* // CHECK10-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP81]], align 8 // CHECK10-NEXT: [[TMP82:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 2 // CHECK10-NEXT: store i32 [[TMP82]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK10-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i64* // CHECK10-NEXT: [[TMP83:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP83]], align 8 // CHECK10-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP83]], 0
// CHECK10-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 2 // CHECK10-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK10-NEXT: store i64 8, i64* [[TMP84]], align 8 // CHECK10-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK10-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 2 // CHECK10-NEXT: store i32 [[SUB28]], i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP85]], align 8 // CHECK10-NEXT: [[TMP84:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 3 // CHECK10-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP84]], 1
// CHECK10-NEXT: [[TMP87:%.*]] = bitcast i8** [[TMP86]] to i32** // CHECK10-NEXT: [[TMP85:%.*]] = zext i32 [[ADD29]] to i64
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP87]], align 8 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP85]])
// CHECK10-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 3 // CHECK10-NEXT: [[TMP86:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118.region_id, i32 4, i8** [[TMP80]], i8** [[TMP81]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to i32** // CHECK10-NEXT: [[TMP87:%.*]] = icmp ne i32 [[TMP86]], 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP89]], align 8 // CHECK10-NEXT: br i1 [[TMP87]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK10-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 3 // CHECK10: omp_offload.failed30:
// CHECK10-NEXT: store i64 [[TMP67]], i64* [[TMP90]], align 8 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118(i64 [[TMP56]], i64 [[TMP58]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS23]], i64 0, i64 3 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK10-NEXT: store i8* null, i8** [[TMP91]], align 8 // CHECK10: omp_offload.cont31:
// CHECK10-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS21]], i32 0, i32 0 // CHECK10-NEXT: [[TMP88:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS22]], i32 0, i32 0 // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP88]])
// CHECK10-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES24]], i32 0, i32 0
// CHECK10-NEXT: [[TMP95:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP95]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[TMP96:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[SUB28:%.*]] = sub nsw i32 [[TMP96]], 0
// CHECK10-NEXT: [[DIV29:%.*]] = sdiv i32 [[SUB28]], 1
// CHECK10-NEXT: [[SUB30:%.*]] = sub nsw i32 [[DIV29]], 1
// CHECK10-NEXT: store i32 [[SUB30]], i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[TMP97:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_27]], align 4
// CHECK10-NEXT: [[ADD31:%.*]] = add nsw i32 [[TMP97]], 1
// CHECK10-NEXT: [[TMP98:%.*]] = zext i32 [[ADD31]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP98]])
// CHECK10-NEXT: [[TMP99:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118.region_id, i32 4, i8** [[TMP92]], i8** [[TMP93]], i64* [[TMP94]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP100:%.*]] = icmp ne i32 [[TMP99]], 0
// CHECK10-NEXT: br i1 [[TMP100]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]]
// CHECK10: omp_offload.failed32:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118(i64 [[TMP64]], i64 [[TMP66]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT33]]
// CHECK10: omp_offload.cont33:
// CHECK10-NEXT: [[TMP101:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP101]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP102:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP89:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP102]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP89]])
// CHECK10-NEXT: [[TMP103:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP90:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP103]] // CHECK10-NEXT: ret i32 [[TMP90]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 481 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV2]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV2]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* [[CONV1]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[CONV1]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 65 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 // CHECK10-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
// CHECK10-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 // CHECK10-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
// CHECK10-NEXT: [[TMP21:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP21:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[CONV7:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV7:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP21]], i32* [[CONV7]], align 4, !llvm.access.group !24 // CHECK10-NEXT: store i32 [[TMP21]], i32* [[CONV7]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !24 // CHECK10-NEXT: [[TMP22:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !24
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64, i32*, i64)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i64 [[TMP1]], i32* [[TMP2]], i64 [[TMP22]]), !llvm.access.group !24 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64, i32*, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP18]], i64 [[TMP20]], i32* [[TMP0]], i64 [[TMP1]], i32* [[TMP2]], i64 [[TMP22]]), !llvm.access.group !24
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]] // CHECK10-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
// CHECK10-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK10-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !24
// CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24 // CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !24
Show All 37 Lines
// CHECK10-NEXT: store i32 [[ADD18]], i32* [[I4]], align 4 // CHECK10-NEXT: store i32 [[ADD18]], i32* [[I4]], align 4
// CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK10: .omp.final.done: // CHECK10: .omp.final.done:
// CHECK10-NEXT: br label [[OMP_PRECOND_END]] // CHECK10-NEXT: br label [[OMP_PRECOND_END]]
// CHECK10: omp.precond.end: // CHECK10: omp.precond.end:
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 135 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK10: omp_offload.failed5: // CHECK10: omp_offload.failed5:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK10: omp_offload.cont6: // CHECK10: omp_offload.cont6:
// CHECK10-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK10-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[M_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[M_CASTED]] to i32*
Show All 13 Lines
// CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]** // CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 8
// CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1 // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK10-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 // CHECK10-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// CHECK10-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK10-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK10: omp_offload.failed11: // CHECK10: omp_offload.failed11:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96(i64 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96(i64 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK10: omp_offload.cont12: // CHECK10: omp_offload.cont12:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !30 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !30
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !30
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !30 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !30
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !30 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !30
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !30
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP31:![0-9]+]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK10: .omp.final.then: // CHECK10: .omp.final.then:
// CHECK10-NEXT: store i32 10, i32* [[I]], align 4 // CHECK10-NEXT: store i32 10, i32* [[I]], align 4
// CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK10: .omp.final.done: // CHECK10: .omp.final.done:
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 31 Lines
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !36 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !36
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !36
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !36 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]), !llvm.access.group !36
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !36 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !36
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !36
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP37:![0-9]+]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK10: .omp.final.then: // CHECK10: .omp.final.then:
// CHECK10-NEXT: store i32 10, i32* [[I]], align 4 // CHECK10-NEXT: store i32 10, i32* [[I]], align 4
// CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK10: .omp.final.done: // CHECK10: .omp.final.done:
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[M_ADDR]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[M_ADDR]] to i32*
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4 // CHECK10-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP3:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP3]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i64)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i64 [[TMP3]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 39 Lines
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[CONV]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV2:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4, !llvm.access.group !42 // CHECK10-NEXT: store i32 [[TMP12]], i32* [[CONV2]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !42 // CHECK10-NEXT: [[TMP13:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8, !llvm.access.group !42
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]), !llvm.access.group !42 // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*, i64)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]], i64 [[TMP13]]), !llvm.access.group !42
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42 // CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !42
// CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42 // CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !42
Show All 26 Lines
// CHECK10-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK10-NEXT: br i1 [[TMP24]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK10: .omp.final.then: // CHECK10: .omp.final.then:
// CHECK10-NEXT: store i32 10, i32* [[I]], align 4 // CHECK10-NEXT: store i32 10, i32* [[I]], align 4
// CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK10-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK10: .omp.final.done: // CHECK10: .omp.final.done:
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 91 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK11-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED16:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS17:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS19:%.*]] = alloca [4 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES21:%.*]] = alloca [4 x i64], align 4 // CHECK11-NEXT: [[_TMP20:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP22:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_21:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: store i32 10, i32* [[M]], align 4 // CHECK11-NEXT: store i32 10, i32* [[M]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK11-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK11-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK11-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK11-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK11-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK11-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK11-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK11-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK11-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK11-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP53]])
// CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK11-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK11: omp_offload.failed14:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK11: omp_offload.cont15:
// CHECK11-NEXT: [[TMP56:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP56]], i32* [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP57:%.*]] = load i32, i32* [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11-NEXT: store i32 [[TMP58]], i32* [[N_CASTED16]], align 4
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[N_CASTED16]], align 4
// CHECK11-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK11-NEXT: [[TMP60:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK11-NEXT: [[TMP61:%.*]] = sext i32 [[TMP60]] to i64
// CHECK11-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK11-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32*
// CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: store i32 [[TMP57]], i32* [[TMP63]], align 4
// CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK11-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK11-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK11-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32*
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP61]]) // CHECK11-NEXT: store i32 [[TMP57]], i32* [[TMP65]], align 4
// CHECK11-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 0
// CHECK11-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK11-NEXT: store i8* null, i8** [[TMP66]], align 4
// CHECK11-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 1
// CHECK11: omp_offload.failed15: // CHECK11-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i32*
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK11-NEXT: store i32 [[TMP59]], i32* [[TMP68]], align 4
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK11-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 1
// CHECK11: omp_offload.cont16: // CHECK11-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i32*
// CHECK11-NEXT: [[TMP64:%.*]] = load i32, i32* [[M]], align 4 // CHECK11-NEXT: store i32 [[TMP59]], i32* [[TMP70]], align 4
// CHECK11-NEXT: store i32 [[TMP64]], i32* [[M_CASTED]], align 4 // CHECK11-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 1
// CHECK11-NEXT: [[TMP65:%.*]] = load i32, i32* [[M_CASTED]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP71]], align 4
// CHECK11-NEXT: [[TMP66:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP66]], i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: [[TMP67:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK11-NEXT: [[TMP68:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP69:%.*]] = sext i32 [[TMP68]] to i64
// CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK11-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[TMP71]], align 4
// CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK11-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32* // CHECK11-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32*
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[TMP73]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP73]], align 4
// CHECK11-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK11-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP74]], align 4 // CHECK11-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32*
// CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP75]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP75]], align 4 // CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 2
// CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP76]], align 4
// CHECK11-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32* // CHECK11-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP67]], i32* [[TMP77]], align 4 // CHECK11-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32**
// CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP78]], align 4
// CHECK11-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32* // CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP67]], i32* [[TMP79]], align 4 // CHECK11-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32**
// CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP80]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP80]], align 4 // CHECK11-NEXT: store i64 [[TMP61]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1 // CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP81]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP81]], align 4
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2 // CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK11-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i32* // CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP83]], align 4 // CHECK11-NEXT: [[TMP84:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[TMP84]], i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK11-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to i32* // CHECK11-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP85]], align 4 // CHECK11-NEXT: [[SUB23:%.*]] = sub nsw i32 [[TMP85]], 0
// CHECK11-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 2 // CHECK11-NEXT: [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
// CHECK11-NEXT: store i64 4, i64* [[TMP86]], align 4 // CHECK11-NEXT: [[SUB25:%.*]] = sub nsw i32 [[DIV24]], 1
// CHECK11-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2 // CHECK11-NEXT: store i32 [[SUB25]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP87]], align 4 // CHECK11-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3 // CHECK11-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP86]], 1
// CHECK11-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to i32** // CHECK11-NEXT: [[TMP87:%.*]] = zext i32 [[ADD26]] to i64
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP89]], align 4 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP87]])
// CHECK11-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3 // CHECK11-NEXT: [[TMP88:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118.region_id, i32 4, i8** [[TMP82]], i8** [[TMP83]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to i32** // CHECK11-NEXT: [[TMP89:%.*]] = icmp ne i32 [[TMP88]], 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP91]], align 4 // CHECK11-NEXT: br i1 [[TMP89]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK11-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 3 // CHECK11: omp_offload.failed27:
// CHECK11-NEXT: store i64 [[TMP69]], i64* [[TMP92]], align 4 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118(i32 [[TMP57]], i32 [[TMP59]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK11-NEXT: store i8* null, i8** [[TMP93]], align 4 // CHECK11: omp_offload.cont28:
// CHECK11-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK11-NEXT: [[TMP90:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP90]])
// CHECK11-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0
// CHECK11-NEXT: [[TMP97:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP97]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP98]], 0
// CHECK11-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK11-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK11-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP99]], 1
// CHECK11-NEXT: [[TMP100:%.*]] = zext i32 [[ADD28]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP100]])
// CHECK11-NEXT: [[TMP101:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118.region_id, i32 4, i8** [[TMP94]], i8** [[TMP95]], i64* [[TMP96]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP102:%.*]] = icmp ne i32 [[TMP101]], 0
// CHECK11-NEXT: br i1 [[TMP102]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK11: omp_offload.failed29:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118(i32 [[TMP65]], i32 [[TMP67]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK11: omp_offload.cont30:
// CHECK11-NEXT: [[TMP103:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP103]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP104:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP91:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP104]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP91]])
// CHECK11-NEXT: [[TMP105:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP92:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP105]] // CHECK11-NEXT: ret i32 [[TMP92]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 466 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[M_ADDR]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[M_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]] // CHECK11-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
// CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !25
// CHECK11-NEXT: store i32 [[TMP19]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25 // CHECK11-NEXT: store i32 [[TMP19]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32 [[TMP1]], i32* [[TMP2]], i32 [[TMP20]]), !llvm.access.group !25 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32, i32*, i32)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32 [[TMP1]], i32* [[TMP2]], i32 [[TMP20]]), !llvm.access.group !25
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
// CHECK11-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 // CHECK11-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25
// CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25
Show All 37 Lines
// CHECK11-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4 // CHECK11-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4
// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK11: .omp.final.done: // CHECK11: .omp.final.done:
// CHECK11-NEXT: br label [[OMP_PRECOND_END]] // CHECK11-NEXT: br label [[OMP_PRECOND_END]]
// CHECK11: omp.precond.end: // CHECK11: omp.precond.end:
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 131 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK11: omp_offload.failed5: // CHECK11: omp_offload.failed5:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK11: omp_offload.cont6: // CHECK11: omp_offload.cont6:
// CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP18]], i32* [[M_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP18]], i32* [[M_CASTED]], align 4
Show All 12 Lines
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK11-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]** // CHECK11-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP29]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK11-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK11-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 // CHECK11-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// CHECK11-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK11-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK11: omp_offload.failed11: // CHECK11: omp_offload.failed11:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96(i32 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96(i32 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK11: omp_offload.cont12: // CHECK11: omp_offload.cont12:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31
// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !31
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !31 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !31
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !31 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !31
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK11-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK11-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK11-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then: // CHECK11: .omp.final.then:
// CHECK11-NEXT: store i32 10, i32* [[I]], align 4 // CHECK11-NEXT: store i32 10, i32* [[I]], align 4
// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK11: .omp.final.done: // CHECK11: .omp.final.done:
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37
// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !37
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !37 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !37
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !37 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !37
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP38:![0-9]+]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP38:![0-9]+]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK11-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK11-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK11-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then: // CHECK11: .omp.final.then:
// CHECK11-NEXT: store i32 10, i32* [[I]], align 4 // CHECK11-NEXT: store i32 10, i32* [[I]], align 4
// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK11: .omp.final.done: // CHECK11: .omp.final.done:
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 79 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: store i32 [[M]], i32* [[M_ADDR]], align 4 // CHECK11-NEXT: store i32 [[M]], i32* [[M_ADDR]], align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[M_ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[M_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP3]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP3]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 35 Lines
// CHECK11-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10 // CHECK11-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10
// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !43
// CHECK11-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43 // CHECK11-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]), !llvm.access.group !43 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]), !llvm.access.group !43
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43
// CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43
Show All 26 Lines
// CHECK11-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK11-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then: // CHECK11: .omp.final.then:
// CHECK11-NEXT: store i32 10, i32* [[I]], align 4 // CHECK11-NEXT: store i32 10, i32* [[I]], align 4
// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK11: .omp.final.done: // CHECK11: .omp.final.done:
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 87 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK12-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED16:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS17:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS18:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [4 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS19:%.*]] = alloca [4 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES21:%.*]] = alloca [4 x i64], align 4 // CHECK12-NEXT: [[_TMP20:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP22:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_21:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: store i32 10, i32* [[M]], align 4 // CHECK12-NEXT: store i32 10, i32* [[M]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK12-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK12-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK12-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK12-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 2), align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK12-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK12-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK12-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK12-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK12-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK12-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK12-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK12-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK12-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK12-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK12-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK12-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP53]])
// CHECK12-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK12-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK12: omp_offload.failed14:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK12: omp_offload.cont15:
// CHECK12-NEXT: [[TMP56:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP56]], i32* [[M_CASTED]], align 4
// CHECK12-NEXT: [[TMP57:%.*]] = load i32, i32* [[M_CASTED]], align 4
// CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12-NEXT: store i32 [[TMP58]], i32* [[N_CASTED16]], align 4
// CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[N_CASTED16]], align 4
// CHECK12-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK12-NEXT: [[TMP60:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK12-NEXT: [[TMP61:%.*]] = sext i32 [[TMP60]] to i64
// CHECK12-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK12-NEXT: [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK12-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32*
// CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: store i32 [[TMP57]], i32* [[TMP63]], align 4
// CHECK12-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK12-NEXT: [[TMP64:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK12-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK12-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32*
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP61]]) // CHECK12-NEXT: store i32 [[TMP57]], i32* [[TMP65]], align 4
// CHECK12-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP66:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 0
// CHECK12-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK12-NEXT: store i8* null, i8** [[TMP66]], align 4
// CHECK12-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK12-NEXT: [[TMP67:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 1
// CHECK12: omp_offload.failed15: // CHECK12-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i32*
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l113(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK12-NEXT: store i32 [[TMP59]], i32* [[TMP68]], align 4
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK12-NEXT: [[TMP69:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 1
// CHECK12: omp_offload.cont16: // CHECK12-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i32*
// CHECK12-NEXT: [[TMP64:%.*]] = load i32, i32* [[M]], align 4 // CHECK12-NEXT: store i32 [[TMP59]], i32* [[TMP70]], align 4
// CHECK12-NEXT: store i32 [[TMP64]], i32* [[M_CASTED]], align 4 // CHECK12-NEXT: [[TMP71:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 1
// CHECK12-NEXT: [[TMP65:%.*]] = load i32, i32* [[M_CASTED]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP71]], align 4
// CHECK12-NEXT: [[TMP66:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP66]], i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: [[TMP67:%.*]] = load i32, i32* [[N_CASTED17]], align 4
// CHECK12-NEXT: [[TMP68:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP69:%.*]] = sext i32 [[TMP68]] to i64
// CHECK12-NEXT: [[TMP70:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0
// CHECK12-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32*
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[TMP71]], align 4
// CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0
// CHECK12-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32* // CHECK12-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32*
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[TMP73]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP73]], align 4
// CHECK12-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK12-NEXT: [[TMP74:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP74]], align 4 // CHECK12-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32*
// CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP75]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP75]], align 4 // CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 2
// CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP76]], align 4
// CHECK12-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32* // CHECK12-NEXT: [[TMP77:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP67]], i32* [[TMP77]], align 4 // CHECK12-NEXT: [[TMP78:%.*]] = bitcast i8** [[TMP77]] to i32**
// CHECK12-NEXT: [[TMP78:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP78]], align 4
// CHECK12-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32* // CHECK12-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP67]], i32* [[TMP79]], align 4 // CHECK12-NEXT: [[TMP80:%.*]] = bitcast i8** [[TMP79]] to i32**
// CHECK12-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP80]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP80]], align 4 // CHECK12-NEXT: store i64 [[TMP61]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 3), align 4
// CHECK12-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1 // CHECK12-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP81]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP81]], align 4
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2 // CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK12-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i32* // CHECK12-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP83]], align 4 // CHECK12-NEXT: [[TMP84:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[TMP84]], i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK12-NEXT: [[TMP85:%.*]] = bitcast i8** [[TMP84]] to i32* // CHECK12-NEXT: [[TMP85:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP85]], align 4 // CHECK12-NEXT: [[SUB23:%.*]] = sub nsw i32 [[TMP85]], 0
// CHECK12-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 2 // CHECK12-NEXT: [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
// CHECK12-NEXT: store i64 4, i64* [[TMP86]], align 4 // CHECK12-NEXT: [[SUB25:%.*]] = sub nsw i32 [[DIV24]], 1
// CHECK12-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2 // CHECK12-NEXT: store i32 [[SUB25]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP87]], align 4 // CHECK12-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 3 // CHECK12-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP86]], 1
// CHECK12-NEXT: [[TMP89:%.*]] = bitcast i8** [[TMP88]] to i32** // CHECK12-NEXT: [[TMP87:%.*]] = zext i32 [[ADD26]] to i64
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP89]], align 4 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP87]])
// CHECK12-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 3 // CHECK12-NEXT: [[TMP88:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118.region_id, i32 4, i8** [[TMP82]], i8** [[TMP83]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP91:%.*]] = bitcast i8** [[TMP90]] to i32** // CHECK12-NEXT: [[TMP89:%.*]] = icmp ne i32 [[TMP88]], 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP91]], align 4 // CHECK12-NEXT: br i1 [[TMP89]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK12-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 3 // CHECK12: omp_offload.failed27:
// CHECK12-NEXT: store i64 [[TMP69]], i64* [[TMP92]], align 4 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118(i32 [[TMP57]], i32 [[TMP59]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 3 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK12-NEXT: store i8* null, i8** [[TMP93]], align 4 // CHECK12: omp_offload.cont28:
// CHECK12-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK12-NEXT: [[TMP90:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP90]])
// CHECK12-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0
// CHECK12-NEXT: [[TMP97:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP97]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: [[TMP98:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP98]], 0
// CHECK12-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK12-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK12-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[TMP99:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP99]], 1
// CHECK12-NEXT: [[TMP100:%.*]] = zext i32 [[ADD28]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP100]])
// CHECK12-NEXT: [[TMP101:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118.region_id, i32 4, i8** [[TMP94]], i8** [[TMP95]], i64* [[TMP96]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP102:%.*]] = icmp ne i32 [[TMP101]], 0
// CHECK12-NEXT: br i1 [[TMP102]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK12: omp_offload.failed29:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l118(i32 [[TMP65]], i32 [[TMP67]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK12: omp_offload.cont30:
// CHECK12-NEXT: [[TMP103:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP103]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP104:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP91:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP104]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP91]])
// CHECK12-NEXT: [[TMP105:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP92:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP105]] // CHECK12-NEXT: ret i32 [[TMP92]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 466 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 // CHECK12-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[M_ADDR]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[M_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]] // CHECK12-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
// CHECK12-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !25
// CHECK12-NEXT: store i32 [[TMP19]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25 // CHECK12-NEXT: store i32 [[TMP19]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !25
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32, i32*, i32)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32 [[TMP1]], i32* [[TMP2]], i32 [[TMP20]]), !llvm.access.group !25 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32, i32*, i32)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32 [[TMP17]], i32 [[TMP18]], i32* [[TMP0]], i32 [[TMP1]], i32* [[TMP2]], i32 [[TMP20]]), !llvm.access.group !25
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] // CHECK12-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
// CHECK12-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25 // CHECK12-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !25
// CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !25
Show All 37 Lines
// CHECK12-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4 // CHECK12-NEXT: store i32 [[ADD17]], i32* [[I4]], align 4
// CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK12: .omp.final.done: // CHECK12: .omp.final.done:
// CHECK12-NEXT: br label [[OMP_PRECOND_END]] // CHECK12-NEXT: br label [[OMP_PRECOND_END]]
// CHECK12: omp.precond.end: // CHECK12: omp.precond.end:
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..6 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 131 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.10, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.12, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.13, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.13, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.14, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.16, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.17, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK12: omp_offload.failed5: // CHECK12: omp_offload.failed5:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK12: omp_offload.cont6: // CHECK12: omp_offload.cont6:
// CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP18]], i32* [[M_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP18]], i32* [[M_CASTED]], align 4
Show All 12 Lines
// CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK12-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]** // CHECK12-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP28]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1 // CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP29]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.17, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.18, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) // CHECK12-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96.region_id, i32 2, i8** [[TMP30]], i8** [[TMP31]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.20, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.21, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK12-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 // CHECK12-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// CHECK12-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK12-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK12: omp_offload.failed11: // CHECK12: omp_offload.failed11:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96(i32 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l96(i32 [[TMP19]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK12: omp_offload.cont12: // CHECK12: omp_offload.cont12:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l86
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..8 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31
// CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !31
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !31
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !31 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !31
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !31 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !31
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !31
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP32:![0-9]+]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK12-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK12-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK12-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK12: .omp.final.then: // CHECK12: .omp.final.then:
// CHECK12-NEXT: store i32 10, i32* [[I]], align 4 // CHECK12-NEXT: store i32 10, i32* [[I]], align 4
// CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK12: .omp.final.done: // CHECK12: .omp.final.done:
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..9 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l91
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..11 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..14
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
Show All 29 Lines
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37
// CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK12-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !37
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !37
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !37 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]), !llvm.access.group !37
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !37 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !37
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !37
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP38:![0-9]+]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP38:![0-9]+]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK12-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK12-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK12-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK12: .omp.final.then: // CHECK12: .omp.final.then:
// CHECK12-NEXT: store i32 10, i32* [[I]], align 4 // CHECK12-NEXT: store i32 10, i32* [[I]], align 4
// CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK12: .omp.final.done: // CHECK12: .omp.final.done:
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..12 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 79 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: store i32 [[M]], i32* [[M_ADDR]], align 4 // CHECK12-NEXT: store i32 [[M]], i32* [[M_ADDR]], align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[M_ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32, i32* [[M_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..15 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP3]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*, i32)* @.omp_outlined..18 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]], i32 [[TMP3]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..15 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..18
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
Show All 35 Lines
// CHECK12-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10 // CHECK12-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10
// CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__ADDR]], align 4, !llvm.access.group !43
// CHECK12-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43 // CHECK12-NEXT: store i32 [[TMP10]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4, !llvm.access.group !43
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]), !llvm.access.group !43 // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*, i32)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]], i32 [[TMP11]]), !llvm.access.group !43
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43 // CHECK12-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !43
// CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43 // CHECK12-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4, !llvm.access.group !43
Show All 26 Lines
// CHECK12-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK12-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK12: .omp.final.then: // CHECK12: .omp.final.then:
// CHECK12-NEXT: store i32 10, i32* [[I]], align 4 // CHECK12-NEXT: store i32 10, i32* [[I]], align 4
// CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK12-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK12: .omp.final.done: // CHECK12: .omp.final.done:
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..16 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 1,346 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_distribute_parallel_for_simd_schedule_codegen.cpp

This file has a very large number of changes (5,360 lines). Show File Contents

clang/test/OpenMP/teams_distribute_simd_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 2,363 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK9-NEXT: ret i32 [[TMP33]] // CHECK9-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 126 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 // CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK10-NEXT: ret i32 [[TMP33]] // CHECK10-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 126 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK11-NEXT: ret i32 [[TMP33]] // CHECK11-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 124 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 // CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK12-NEXT: ret i32 [[TMP33]] // CHECK12-NEXT: ret i32 [[TMP29]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 433 Lines • ▼ Show 20 Lines
// //
// CHECK17-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv // CHECK17-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK17-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { // CHECK17-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK17-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 // CHECK17-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK17-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 // CHECK17-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK17-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 // CHECK17-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1
// CHECK17-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* // CHECK17-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8*
// CHECK17-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* // CHECK17-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8*
// CHECK17-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK17-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK17-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK17-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK17-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK17-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK17-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK17-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK17-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS** // CHECK17-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS**
// CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 8 // CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 8
// CHECK17-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]** // CHECK17-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]**
// CHECK17-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 8 // CHECK17-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 8
// CHECK17-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK17-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), align 8
// CHECK17-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 8 // CHECK17-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK17-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP12]], align 8 // CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK17-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK17-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.SS**
// CHECK17-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.SS** // CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP13]], align 8
// CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP14]], align 8 // CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK17-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK17-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK17-NEXT: store float* [[B]], float** [[TMP15]], align 8
// CHECK17-NEXT: store float* [[B]], float** [[TMP16]], align 8 // CHECK17-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK17-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK17-NEXT: store i64 4, i64* [[TMP17]], align 8 // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.SS**
// CHECK17-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP18]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** // CHECK17-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [123 x i32]**
// CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 8 // CHECK17-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP20]], align 8
// CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK17-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** // CHECK17-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK17-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 8 // CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK17-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: store i64 492, i64* [[TMP23]], align 8
// CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK17-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK17-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) // CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123)
// CHECK17-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK17-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK17-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 // CHECK17-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK17-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK17-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK17-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0
// CHECK17-NEXT: [[TMP30:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK17-NEXT: [[TMP26:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK17-NEXT: ret i32 [[TMP30]] // CHECK17-NEXT: ret i32 [[TMP26]]
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123
// CHECK17-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK17-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 // CHECK17-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
▲ Show 20 Lines • Show All 96 Lines • ▼ Show 20 Lines
// //
// CHECK18-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv // CHECK18-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK18-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { // CHECK18-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 // CHECK18-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK18-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK18-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK18-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 // CHECK18-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK18-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 // CHECK18-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK18-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 // CHECK18-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK18-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK18-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK18-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 // CHECK18-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1
// CHECK18-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* // CHECK18-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8*
// CHECK18-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* // CHECK18-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8*
// CHECK18-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK18-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK18-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK18-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK18-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK18-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK18-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK18-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK18-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS** // CHECK18-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS**
// CHECK18-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 8 // CHECK18-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 8
// CHECK18-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]** // CHECK18-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]**
// CHECK18-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 8 // CHECK18-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 8
// CHECK18-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK18-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), align 8
// CHECK18-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 8 // CHECK18-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK18-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK18-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP12]], align 8 // CHECK18-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK18-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK18-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.SS**
// CHECK18-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.SS** // CHECK18-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP13]], align 8
// CHECK18-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP14]], align 8 // CHECK18-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK18-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK18-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK18-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK18-NEXT: store float* [[B]], float** [[TMP15]], align 8
// CHECK18-NEXT: store float* [[B]], float** [[TMP16]], align 8 // CHECK18-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK18-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK18-NEXT: store i64 4, i64* [[TMP17]], align 8 // CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.SS**
// CHECK18-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK18-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP18]], align 8
// CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** // CHECK18-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [123 x i32]**
// CHECK18-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 8 // CHECK18-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP20]], align 8
// CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK18-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** // CHECK18-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK18-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 8 // CHECK18-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK18-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: store i64 492, i64* [[TMP23]], align 8
// CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK18-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK18-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK18-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) // CHECK18-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123)
// CHECK18-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK18-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK18-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 // CHECK18-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK18-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK18-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK18-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK18-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0 // CHECK18-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0
// CHECK18-NEXT: [[TMP30:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK18-NEXT: [[TMP26:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK18-NEXT: ret i32 [[TMP30]] // CHECK18-NEXT: ret i32 [[TMP26]]
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 // CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123
// CHECK18-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK18-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 // CHECK18-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK18-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 // CHECK18-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK18-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 // CHECK18-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
▲ Show 20 Lines • Show All 96 Lines • ▼ Show 20 Lines
// //
// CHECK19-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv // CHECK19-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK19-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { // CHECK19-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK19-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 // CHECK19-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK19-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 // CHECK19-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK19-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 // CHECK19-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1
// CHECK19-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* // CHECK19-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8*
// CHECK19-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* // CHECK19-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8*
// CHECK19-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK19-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK19-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK19-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK19-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK19-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK19-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK19-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK19-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS** // CHECK19-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS**
// CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 4 // CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 4
// CHECK19-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]** // CHECK19-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]**
// CHECK19-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 4 // CHECK19-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 4
// CHECK19-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK19-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), align 4
// CHECK19-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 4 // CHECK19-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK19-NEXT: store i8* null, i8** [[TMP11]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP12]], align 4 // CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK19-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK19-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.SS**
// CHECK19-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.SS** // CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP13]], align 4
// CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP14]], align 4 // CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK19-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK19-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK19-NEXT: store float* [[B]], float** [[TMP15]], align 4
// CHECK19-NEXT: store float* [[B]], float** [[TMP16]], align 4 // CHECK19-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK19-NEXT: store i8* null, i8** [[TMP16]], align 4
// CHECK19-NEXT: store i64 4, i64* [[TMP17]], align 4 // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK19-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.SS**
// CHECK19-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP18]], align 4
// CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** // CHECK19-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [123 x i32]**
// CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 4 // CHECK19-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP20]], align 4
// CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** // CHECK19-NEXT: store i8* null, i8** [[TMP21]], align 4
// CHECK19-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 4 // CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK19-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: store i64 492, i64* [[TMP23]], align 4
// CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK19-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) // CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123)
// CHECK19-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK19-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK19-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 // CHECK19-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK19-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK19-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK19-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0
// CHECK19-NEXT: [[TMP30:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK19-NEXT: [[TMP26:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK19-NEXT: ret i32 [[TMP30]] // CHECK19-NEXT: ret i32 [[TMP26]]
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123
// CHECK19-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK19-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
▲ Show 20 Lines • Show All 95 Lines • ▼ Show 20 Lines
// //
// CHECK20-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv // CHECK20-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK20-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { // CHECK20-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 // CHECK20-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK20-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK20-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 // CHECK20-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK20-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 // CHECK20-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK20-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 // CHECK20-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK20-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK20-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK20-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 // CHECK20-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1
// CHECK20-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* // CHECK20-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8*
// CHECK20-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* // CHECK20-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8*
// CHECK20-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK20-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK20-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK20-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK20-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK20-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK20-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK20-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK20-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS** // CHECK20-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS**
// CHECK20-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 4 // CHECK20-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 4
// CHECK20-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]** // CHECK20-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]**
// CHECK20-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 4 // CHECK20-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 4
// CHECK20-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK20-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), align 4
// CHECK20-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 4 // CHECK20-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK20-NEXT: store i8* null, i8** [[TMP11]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP12]], align 4 // CHECK20-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK20-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK20-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.SS**
// CHECK20-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.SS** // CHECK20-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP13]], align 4
// CHECK20-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP14]], align 4 // CHECK20-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK20-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK20-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK20-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK20-NEXT: store float* [[B]], float** [[TMP15]], align 4
// CHECK20-NEXT: store float* [[B]], float** [[TMP16]], align 4 // CHECK20-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK20-NEXT: store i8* null, i8** [[TMP16]], align 4
// CHECK20-NEXT: store i64 4, i64* [[TMP17]], align 4 // CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK20-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.SS**
// CHECK20-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK20-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP18]], align 4
// CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** // CHECK20-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [123 x i32]**
// CHECK20-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 4 // CHECK20-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP20]], align 4
// CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** // CHECK20-NEXT: store i8* null, i8** [[TMP21]], align 4
// CHECK20-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 4 // CHECK20-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK20-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: store i64 492, i64* [[TMP23]], align 4
// CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK20-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK20-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK20-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) // CHECK20-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123)
// CHECK20-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK20-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK20-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 // CHECK20-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK20-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK20-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK20-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK20-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0 // CHECK20-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0
// CHECK20-NEXT: [[TMP30:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK20-NEXT: [[TMP26:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK20-NEXT: ret i32 [[TMP30]] // CHECK20-NEXT: ret i32 [[TMP26]]
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 // CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123
// CHECK20-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK20-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 // CHECK20-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK20-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 // CHECK20-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
▲ Show 20 Lines • Show All 95 Lines • ▼ Show 20 Lines
// //
// CHECK21-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv // CHECK21-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK21-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { // CHECK21-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
// CHECK21-NEXT: entry: // CHECK21-NEXT: entry:
// CHECK21-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 // CHECK21-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK21-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK21-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK21-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK21-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK21-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK21-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK21-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK21-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK21-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK21-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 // CHECK21-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK21-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 // CHECK21-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK21-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 // CHECK21-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK21-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK21-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK21-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 // CHECK21-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1
// CHECK21-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* // CHECK21-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8*
// CHECK21-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* // CHECK21-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8*
// CHECK21-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK21-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK21-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK21-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK21-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK21-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK21-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK21-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK21-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK21-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK21-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS** // CHECK21-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS**
// CHECK21-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 8 // CHECK21-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 8
// CHECK21-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK21-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK21-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]** // CHECK21-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]**
// CHECK21-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 8 // CHECK21-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 8
// CHECK21-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK21-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), align 8
// CHECK21-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 8 // CHECK21-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK21-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK21-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK21-NEXT: store i8* null, i8** [[TMP12]], align 8 // CHECK21-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK21-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK21-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.SS**
// CHECK21-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.SS** // CHECK21-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP13]], align 8
// CHECK21-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP14]], align 8 // CHECK21-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK21-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK21-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK21-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK21-NEXT: store float* [[B]], float** [[TMP15]], align 8
// CHECK21-NEXT: store float* [[B]], float** [[TMP16]], align 8 // CHECK21-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK21-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK21-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK21-NEXT: store i64 4, i64* [[TMP17]], align 8 // CHECK21-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK21-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK21-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.SS**
// CHECK21-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK21-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP18]], align 8
// CHECK21-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK21-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK21-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** // CHECK21-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [123 x i32]**
// CHECK21-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 8 // CHECK21-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP20]], align 8
// CHECK21-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK21-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK21-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** // CHECK21-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK21-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 8 // CHECK21-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK21-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK21-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK21-NEXT: store i64 492, i64* [[TMP23]], align 8
// CHECK21-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK21-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK21-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK21-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK21-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK21-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) // CHECK21-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123)
// CHECK21-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK21-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK21-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 // CHECK21-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK21-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK21-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK21: omp_offload.failed: // CHECK21: omp_offload.failed:
// CHECK21-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK21-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK21-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK21-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK21: omp_offload.cont: // CHECK21: omp_offload.cont:
// CHECK21-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK21-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK21-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0 // CHECK21-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0
// CHECK21-NEXT: [[TMP30:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK21-NEXT: [[TMP26:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK21-NEXT: ret i32 [[TMP30]] // CHECK21-NEXT: ret i32 [[TMP26]]
// //
// //
// CHECK21-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 // CHECK21-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123
// CHECK21-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK21-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK21-NEXT: entry: // CHECK21-NEXT: entry:
// CHECK21-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 // CHECK21-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK21-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK21-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK21-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK21-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 148 Lines • ▼ Show 20 Lines
// //
// CHECK22-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv // CHECK22-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK22-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { // CHECK22-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
// CHECK22-NEXT: entry: // CHECK22-NEXT: entry:
// CHECK22-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 // CHECK22-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK22-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK22-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK22-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK22-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK22-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK22-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK22-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK22-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK22-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK22-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 // CHECK22-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK22-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 // CHECK22-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK22-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 // CHECK22-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK22-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK22-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK22-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 // CHECK22-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1
// CHECK22-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* // CHECK22-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8*
// CHECK22-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* // CHECK22-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8*
// CHECK22-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK22-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK22-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK22-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK22-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK22-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK22-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK22-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK22-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK22-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK22-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS** // CHECK22-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS**
// CHECK22-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 8 // CHECK22-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 8
// CHECK22-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK22-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK22-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]** // CHECK22-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]**
// CHECK22-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 8 // CHECK22-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 8
// CHECK22-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK22-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), align 8
// CHECK22-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 8 // CHECK22-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK22-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK22-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK22-NEXT: store i8* null, i8** [[TMP12]], align 8 // CHECK22-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK22-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK22-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.SS**
// CHECK22-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.SS** // CHECK22-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP13]], align 8
// CHECK22-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP14]], align 8 // CHECK22-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK22-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK22-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK22-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK22-NEXT: store float* [[B]], float** [[TMP15]], align 8
// CHECK22-NEXT: store float* [[B]], float** [[TMP16]], align 8 // CHECK22-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK22-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK22-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK22-NEXT: store i64 4, i64* [[TMP17]], align 8 // CHECK22-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK22-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK22-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.SS**
// CHECK22-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK22-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP18]], align 8
// CHECK22-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK22-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK22-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** // CHECK22-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [123 x i32]**
// CHECK22-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 8 // CHECK22-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP20]], align 8
// CHECK22-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK22-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK22-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** // CHECK22-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK22-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 8 // CHECK22-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK22-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK22-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK22-NEXT: store i64 492, i64* [[TMP23]], align 8
// CHECK22-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK22-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK22-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK22-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK22-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK22-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) // CHECK22-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123)
// CHECK22-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK22-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK22-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 // CHECK22-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK22-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK22-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK22: omp_offload.failed: // CHECK22: omp_offload.failed:
// CHECK22-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK22-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK22-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK22-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK22: omp_offload.cont: // CHECK22: omp_offload.cont:
// CHECK22-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK22-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK22-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0 // CHECK22-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0
// CHECK22-NEXT: [[TMP30:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK22-NEXT: [[TMP26:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK22-NEXT: ret i32 [[TMP30]] // CHECK22-NEXT: ret i32 [[TMP26]]
// //
// //
// CHECK22-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 // CHECK22-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123
// CHECK22-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK22-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK22-NEXT: entry: // CHECK22-NEXT: entry:
// CHECK22-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 // CHECK22-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK22-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK22-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK22-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK22-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 148 Lines • ▼ Show 20 Lines
// //
// CHECK23-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv // CHECK23-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK23-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { // CHECK23-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
// CHECK23-NEXT: entry: // CHECK23-NEXT: entry:
// CHECK23-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 // CHECK23-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK23-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK23-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK23-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK23-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK23-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK23-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK23-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK23-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK23-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK23-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 // CHECK23-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK23-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 // CHECK23-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK23-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 // CHECK23-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK23-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK23-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK23-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 // CHECK23-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1
// CHECK23-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* // CHECK23-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8*
// CHECK23-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* // CHECK23-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8*
// CHECK23-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK23-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK23-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK23-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK23-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK23-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK23-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK23-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK23-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK23-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK23-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS** // CHECK23-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS**
// CHECK23-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 4 // CHECK23-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 4
// CHECK23-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK23-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK23-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]** // CHECK23-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]**
// CHECK23-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 4 // CHECK23-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 4
// CHECK23-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK23-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), align 4
// CHECK23-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 4 // CHECK23-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK23-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK23-NEXT: store i8* null, i8** [[TMP11]], align 4
// CHECK23-NEXT: store i8* null, i8** [[TMP12]], align 4 // CHECK23-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK23-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK23-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.SS**
// CHECK23-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.SS** // CHECK23-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP13]], align 4
// CHECK23-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP14]], align 4 // CHECK23-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK23-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK23-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK23-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK23-NEXT: store float* [[B]], float** [[TMP15]], align 4
// CHECK23-NEXT: store float* [[B]], float** [[TMP16]], align 4 // CHECK23-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK23-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK23-NEXT: store i8* null, i8** [[TMP16]], align 4
// CHECK23-NEXT: store i64 4, i64* [[TMP17]], align 4 // CHECK23-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK23-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK23-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.SS**
// CHECK23-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK23-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP18]], align 4
// CHECK23-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK23-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK23-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** // CHECK23-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [123 x i32]**
// CHECK23-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 4 // CHECK23-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP20]], align 4
// CHECK23-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK23-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK23-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** // CHECK23-NEXT: store i8* null, i8** [[TMP21]], align 4
// CHECK23-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 4 // CHECK23-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK23-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK23-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK23-NEXT: store i64 492, i64* [[TMP23]], align 4
// CHECK23-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK23-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK23-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK23-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK23-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK23-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) // CHECK23-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123)
// CHECK23-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK23-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK23-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 // CHECK23-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK23-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK23-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK23: omp_offload.failed: // CHECK23: omp_offload.failed:
// CHECK23-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK23-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK23-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK23-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK23: omp_offload.cont: // CHECK23: omp_offload.cont:
// CHECK23-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK23-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK23-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0 // CHECK23-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0
// CHECK23-NEXT: [[TMP30:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK23-NEXT: [[TMP26:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK23-NEXT: ret i32 [[TMP30]] // CHECK23-NEXT: ret i32 [[TMP26]]
// //
// //
// CHECK23-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 // CHECK23-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123
// CHECK23-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK23-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK23-NEXT: entry: // CHECK23-NEXT: entry:
// CHECK23-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 // CHECK23-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK23-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK23-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK23-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK23-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 146 Lines • ▼ Show 20 Lines
// //
// CHECK24-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv // CHECK24-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK24-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { // CHECK24-SAME: (%struct.SS* nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
// CHECK24-NEXT: entry: // CHECK24-NEXT: entry:
// CHECK24-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 // CHECK24-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK24-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK24-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK24-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK24-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK24-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK24-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK24-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK24-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK24-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK24-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 // CHECK24-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK24-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 // CHECK24-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK24-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1 // CHECK24-NEXT: [[B:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK24-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK24-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK24-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1 // CHECK24-NEXT: [[TMP0:%.*]] = getelementptr float, float* [[B]], i32 1
// CHECK24-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8* // CHECK24-NEXT: [[TMP1:%.*]] = bitcast [123 x i32]* [[A]] to i8*
// CHECK24-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8* // CHECK24-NEXT: [[TMP2:%.*]] = bitcast float* [[TMP0]] to i8*
// CHECK24-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64 // CHECK24-NEXT: [[TMP3:%.*]] = ptrtoint i8* [[TMP2]] to i64
// CHECK24-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64 // CHECK24-NEXT: [[TMP4:%.*]] = ptrtoint i8* [[TMP1]] to i64
// CHECK24-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]] // CHECK24-NEXT: [[TMP5:%.*]] = sub i64 [[TMP3]], [[TMP4]]
// CHECK24-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK24-NEXT: [[TMP6:%.*]] = sdiv exact i64 [[TMP5]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK24-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK24-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK24-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS** // CHECK24-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.SS**
// CHECK24-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 4 // CHECK24-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP8]], align 4
// CHECK24-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK24-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK24-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]** // CHECK24-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [123 x i32]**
// CHECK24-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 4 // CHECK24-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP10]], align 4
// CHECK24-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK24-NEXT: store i64 [[TMP6]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), align 4
// CHECK24-NEXT: store i64 [[TMP6]], i64* [[TMP11]], align 4 // CHECK24-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK24-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK24-NEXT: store i8* null, i8** [[TMP11]], align 4
// CHECK24-NEXT: store i8* null, i8** [[TMP12]], align 4 // CHECK24-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK24-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK24-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %struct.SS**
// CHECK24-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to %struct.SS** // CHECK24-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP13]], align 4
// CHECK24-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP14]], align 4 // CHECK24-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK24-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK24-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to float**
// CHECK24-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK24-NEXT: store float* [[B]], float** [[TMP15]], align 4
// CHECK24-NEXT: store float* [[B]], float** [[TMP16]], align 4 // CHECK24-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK24-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK24-NEXT: store i8* null, i8** [[TMP16]], align 4
// CHECK24-NEXT: store i64 4, i64* [[TMP17]], align 4 // CHECK24-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK24-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK24-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.SS**
// CHECK24-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK24-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP18]], align 4
// CHECK24-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK24-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK24-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.SS** // CHECK24-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [123 x i32]**
// CHECK24-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP20]], align 4 // CHECK24-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP20]], align 4
// CHECK24-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK24-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK24-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to [123 x i32]** // CHECK24-NEXT: store i8* null, i8** [[TMP21]], align 4
// CHECK24-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP22]], align 4 // CHECK24-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK24-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK24-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK24-NEXT: store i64 492, i64* [[TMP23]], align 4
// CHECK24-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK24-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK24-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK24-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK24-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK24-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) // CHECK24-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123)
// CHECK24-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP25]], i8** [[TMP26]], i64* [[TMP27]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK24-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, i32 3, i8** [[TMP22]], i8** [[TMP23]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK24-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0 // CHECK24-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK24-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK24-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK24: omp_offload.failed: // CHECK24: omp_offload.failed:
// CHECK24-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK24-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK24-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK24-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK24: omp_offload.cont: // CHECK24: omp_offload.cont:
// CHECK24-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 // CHECK24-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK24-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0 // CHECK24-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0
// CHECK24-NEXT: [[TMP30:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 // CHECK24-NEXT: [[TMP26:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK24-NEXT: ret i32 [[TMP30]] // CHECK24-NEXT: ret i32 [[TMP26]]
// //
// //
// CHECK24-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 // CHECK24-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123
// CHECK24-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK24-SAME: (%struct.SS* [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK24-NEXT: entry: // CHECK24-NEXT: entry:
// CHECK24-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 // CHECK24-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK24-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 // CHECK24-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
// CHECK24-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK24-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 770 Lines • ▼ Show 20 Lines
// CHECK33-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK33-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK33-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK33-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK33-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK33-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK33-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK33-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK33-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK33-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK33-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK33-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK33-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK33-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK33-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK33-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK33-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK33-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK33-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK33-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK33-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK33-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK33-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK33-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK33-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK33-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK33-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK33-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK33-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK33-NEXT: store i32 100, i32* [[N]], align 4 // CHECK33-NEXT: store i32 100, i32* [[N]], align 4
// CHECK33-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK33-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK33-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK33-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK33-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK33-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK33-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK33-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK33-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK33-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK33-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK33-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK33-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK33-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK33-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK33-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK33-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK33-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK33-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK33-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK33-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK33-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK33-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK33-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK33-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK33-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK33-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK33-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK33-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK33-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK33-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK33-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK33-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK33-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK33-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK33-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK33-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK33-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK33-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK33-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK33-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK33-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK33-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK33-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK33-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK33-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK33-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK33-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK33-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK33-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK33-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK33-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK33-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK33-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK33-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK33-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK33-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK33-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK33-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK33-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK33-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK33-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK33-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK33-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK33-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK33-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK33-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK33-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK33-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK33-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK33-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK33-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK33-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK33-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK33-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK33-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK33-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK33-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK33-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK33-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK33-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK33-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK33-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK33-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK33-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK33-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK33-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK33-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK33-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK33-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK33-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK33-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK33-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK33-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK33-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK33-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK33-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK33-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK33-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK33-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK33-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK33-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK33-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK33-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK33-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK33-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK33-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK33-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK33-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK33-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK33: omp_offload.failed: // CHECK33: omp_offload.failed:
// CHECK33-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK33-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK33-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK33-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK33: omp_offload.cont: // CHECK33: omp_offload.cont:
// CHECK33-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK33-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK33-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP33]]) // CHECK33-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP29]])
// CHECK33-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK33-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK33-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK33-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK33-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK33-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK33-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK33-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK33-NEXT: ret i32 [[TMP35]] // CHECK33-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK33-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 // CHECK33-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192
// CHECK33-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK33-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK33-NEXT: entry: // CHECK33-NEXT: entry:
// CHECK33-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK33-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK33-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK33-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK33-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK33-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 161 Lines • ▼ Show 20 Lines
// CHECK33-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK33-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK33-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 // CHECK33-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8
// CHECK33-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK33-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK33-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK33-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK33-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK33-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK33-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK33-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK33-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK33-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK33-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK33-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK33-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) // CHECK33-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1)
// CHECK33-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK33-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK33-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK33-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK33: omp_offload.failed: // CHECK33: omp_offload.failed:
// CHECK33-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK33-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK33-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK33-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK33: omp_offload.cont: // CHECK33: omp_offload.cont:
// CHECK33-NEXT: ret i32 0 // CHECK33-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 107 Lines • ▼ Show 20 Lines
// CHECK34-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK34-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK34-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK34-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK34-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK34-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK34-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK34-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK34-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK34-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK34-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK34-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK34-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK34-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK34-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK34-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK34-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK34-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK34-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK34-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK34-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK34-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK34-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK34-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK34-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK34-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK34-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK34-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK34-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK34-NEXT: store i32 100, i32* [[N]], align 4 // CHECK34-NEXT: store i32 100, i32* [[N]], align 4
// CHECK34-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK34-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK34-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK34-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK34-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK34-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK34-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK34-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK34-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK34-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK34-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK34-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK34-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK34-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK34-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK34-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK34-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK34-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK34-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK34-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK34-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK34-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK34-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK34-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK34-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK34-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK34-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK34-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK34-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK34-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK34-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK34-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK34-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK34-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK34-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK34-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK34-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK34-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK34-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK34-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK34-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK34-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK34-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK34-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK34-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK34-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK34-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK34-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK34-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK34-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK34-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK34-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK34-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK34-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK34-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK34-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK34-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK34-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK34-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK34-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK34-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK34-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK34-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK34-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK34-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK34-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK34-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK34-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK34-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK34-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK34-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK34-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK34-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK34-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK34-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK34-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK34-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK34-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK34-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK34-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK34-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK34-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK34-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK34-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK34-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK34-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK34-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK34-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK34-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK34-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK34-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK34-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK34-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK34-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK34-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK34-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK34-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK34-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK34-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK34-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK34-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK34-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK34-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK34-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK34-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK34-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK34-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK34-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK34-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK34-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK34: omp_offload.failed: // CHECK34: omp_offload.failed:
// CHECK34-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK34-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK34-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK34-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK34: omp_offload.cont: // CHECK34: omp_offload.cont:
// CHECK34-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK34-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK34-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP33]]) // CHECK34-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP29]])
// CHECK34-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK34-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK34-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK34-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK34-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK34-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK34-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK34-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK34-NEXT: ret i32 [[TMP35]] // CHECK34-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK34-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 // CHECK34-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192
// CHECK34-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK34-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK34-NEXT: entry: // CHECK34-NEXT: entry:
// CHECK34-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK34-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK34-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK34-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK34-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK34-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 161 Lines • ▼ Show 20 Lines
// CHECK34-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK34-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK34-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 // CHECK34-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8
// CHECK34-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK34-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK34-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK34-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK34-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK34-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK34-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK34-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK34-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK34-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK34-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK34-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK34-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) // CHECK34-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1)
// CHECK34-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK34-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK34-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK34-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK34: omp_offload.failed: // CHECK34: omp_offload.failed:
// CHECK34-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK34-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK34-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK34-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK34: omp_offload.cont: // CHECK34: omp_offload.cont:
// CHECK34-NEXT: ret i32 0 // CHECK34-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 107 Lines • ▼ Show 20 Lines
// CHECK35-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK35-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK35-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK35-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK35-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK35-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK35-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK35-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK35-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK35-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK35-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK35-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK35-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK35-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK35-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK35-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK35-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK35-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK35-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK35-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK35-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK35-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK35-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK35-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK35-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK35-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK35-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK35-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK35-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK35-NEXT: store i32 100, i32* [[N]], align 4 // CHECK35-NEXT: store i32 100, i32* [[N]], align 4
// CHECK35-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK35-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK35-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK35-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK35-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK35-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK35-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK35-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK35-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK35-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK35-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK35-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK35-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK35-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK35-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK35-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK35-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK35-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK35-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK35-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK35-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK35-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK35-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK35-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK35-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK35-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK35-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK35-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK35-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK35-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK35-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK35-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK35-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK35-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK35-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK35-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK35-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK35-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK35-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK35-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK35-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK35-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK35-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK35-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK35-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK35-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK35-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK35-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK35-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK35-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK35-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK35-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK35-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK35-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK35-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK35-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK35-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK35-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK35-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK35-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK35-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK35-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK35-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK35-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK35-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK35-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK35-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK35-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK35-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK35-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK35-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK35-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK35-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK35-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK35-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK35-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK35-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK35-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK35-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK35-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK35-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK35-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK35-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK35-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK35-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK35-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK35-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK35-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK35-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK35-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK35-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK35-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK35-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK35-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK35-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK35-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK35-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK35-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK35-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK35-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK35-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK35-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK35-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK35-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK35-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK35-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK35-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK35: omp_offload.failed: // CHECK35: omp_offload.failed:
// CHECK35-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK35-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK35-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK35-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK35: omp_offload.cont: // CHECK35: omp_offload.cont:
// CHECK35-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK35-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK35-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP33]]) // CHECK35-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP29]])
// CHECK35-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK35-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK35-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK35-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK35-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK35-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK35-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK35-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK35-NEXT: ret i32 [[TMP35]] // CHECK35-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK35-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 // CHECK35-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192
// CHECK35-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK35-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK35-NEXT: entry: // CHECK35-NEXT: entry:
// CHECK35-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK35-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK35-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK35-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK35-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK35-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 157 Lines • ▼ Show 20 Lines
// CHECK35-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK35-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK35-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 // CHECK35-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4
// CHECK35-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK35-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK35-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK35-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK35-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK35-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK35-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK35-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK35-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK35-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK35-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK35-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK35-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) // CHECK35-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1)
// CHECK35-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK35-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK35-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK35-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK35: omp_offload.failed: // CHECK35: omp_offload.failed:
// CHECK35-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK35-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK35-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK35-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK35: omp_offload.cont: // CHECK35: omp_offload.cont:
// CHECK35-NEXT: ret i32 0 // CHECK35-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 104 Lines • ▼ Show 20 Lines
// CHECK36-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK36-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK36-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK36-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK36-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK36-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK36-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK36-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK36-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK36-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK36-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK36-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK36-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK36-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK36-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK36-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK36-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK36-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK36-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK36-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK36-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK36-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK36-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK36-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK36-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK36-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK36-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK36-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK36-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK36-NEXT: store i32 100, i32* [[N]], align 4 // CHECK36-NEXT: store i32 100, i32* [[N]], align 4
// CHECK36-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK36-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK36-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK36-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK36-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK36-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK36-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK36-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK36-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK36-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK36-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK36-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK36-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK36-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK36-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK36-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK36-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK36-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK36-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK36-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK36-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK36-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK36-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK36-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK36-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK36-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK36-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK36-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK36-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK36-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK36-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK36-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK36-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK36-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK36-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK36-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK36-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK36-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK36-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK36-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK36-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK36-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK36-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK36-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK36-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK36-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK36-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK36-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK36-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK36-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK36-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK36-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK36-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK36-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK36-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK36-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK36-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK36-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK36-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK36-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK36-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK36-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK36-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK36-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK36-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK36-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK36-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK36-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK36-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK36-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK36-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK36-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK36-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK36-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK36-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK36-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK36-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK36-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK36-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK36-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK36-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK36-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK36-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK36-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK36-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK36-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK36-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK36-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK36-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK36-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK36-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK36-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK36-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK36-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK36-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK36-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK36-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK36-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK36-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK36-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK36-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK36-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK36-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK36-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK36-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK36-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK36-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK36-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK36: omp_offload.failed: // CHECK36: omp_offload.failed:
// CHECK36-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK36-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK36-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK36-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK36: omp_offload.cont: // CHECK36: omp_offload.cont:
// CHECK36-NEXT: [[TMP33:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK36-NEXT: [[TMP29:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK36-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP33]]) // CHECK36-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP29]])
// CHECK36-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK36-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK36-NEXT: [[TMP34:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK36-NEXT: [[TMP30:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK36-NEXT: call void @llvm.stackrestore(i8* [[TMP34]]) // CHECK36-NEXT: call void @llvm.stackrestore(i8* [[TMP30]])
// CHECK36-NEXT: [[TMP35:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK36-NEXT: [[TMP31:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK36-NEXT: ret i32 [[TMP35]] // CHECK36-NEXT: ret i32 [[TMP31]]
// //
// //
// CHECK36-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 // CHECK36-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192
// CHECK36-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK36-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK36-NEXT: entry: // CHECK36-NEXT: entry:
// CHECK36-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK36-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK36-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK36-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK36-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK36-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 157 Lines • ▼ Show 20 Lines
// CHECK36-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK36-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK36-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 // CHECK36-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4
// CHECK36-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK36-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK36-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK36-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK36-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK36-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK36-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK36-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK36-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK36-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK36-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK36-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK36-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) // CHECK36-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1)
// CHECK36-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK36-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK36-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK36-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK36: omp_offload.failed: // CHECK36: omp_offload.failed:
// CHECK36-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK36-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK36-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK36-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK36: omp_offload.cont: // CHECK36: omp_offload.cont:
// CHECK36-NEXT: ret i32 0 // CHECK36-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 105 Lines • ▼ Show 20 Lines
// CHECK37-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK37-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK37-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK37-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK37-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK37-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK37-NEXT: [[ARGC_CASTED:%.*]] = alloca i64, align 8 // CHECK37-NEXT: [[ARGC_CASTED:%.*]] = alloca i64, align 8
// CHECK37-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK37-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK37-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK37-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK37-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK37-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK37-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 // CHECK37-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
// CHECK37-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 8
// CHECK37-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK37-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK37-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK37-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK37-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK37-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK37-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK37-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK37-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK37-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK37-NEXT: store i32 100, i32* [[N]], align 4 // CHECK37-NEXT: store i32 100, i32* [[N]], align 4
// CHECK37-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK37-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
Show All 12 Lines
// CHECK37-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK37-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK37-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK37-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK37-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK37-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK37-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK37-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK37-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK37-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK37-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK37-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK37-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64* // CHECK37-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64*
// CHECK37-NEXT: store i64 [[TMP4]], i64* [[TMP11]], align 8 // CHECK37-NEXT: store i64 [[TMP4]], i64* [[TMP11]], align 8
// CHECK37-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK37-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK37-NEXT: store i64 4, i64* [[TMP12]], align 8 // CHECK37-NEXT: store i8* null, i8** [[TMP12]], align 8
// CHECK37-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK37-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK37-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK37-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK37-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK37-NEXT: store i64 [[TMP6]], i64* [[TMP14]], align 8
// CHECK37-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK37-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK37-NEXT: store i64 [[TMP6]], i64* [[TMP15]], align 8 // CHECK37-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK37-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK37-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK37-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK37-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK37-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK37-NEXT: store i8* null, i8** [[TMP17]], align 8
// CHECK37-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK37-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK37-NEXT: store i64 4, i64* [[TMP18]], align 8 // CHECK37-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64*
// CHECK37-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK37-NEXT: store i64 [[TMP1]], i64* [[TMP19]], align 8
// CHECK37-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK37-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK37-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK37-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK37-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK37-NEXT: store i64 [[TMP1]], i64* [[TMP21]], align 8 // CHECK37-NEXT: store i64 [[TMP1]], i64* [[TMP21]], align 8
// CHECK37-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK37-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK37-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK37-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK37-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8 // CHECK37-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK37-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK37-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32**
// CHECK37-NEXT: store i64 8, i64* [[TMP24]], align 8 // CHECK37-NEXT: store i32* [[VLA]], i32** [[TMP24]], align 8
// CHECK37-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK37-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK37-NEXT: store i8* null, i8** [[TMP25]], align 8 // CHECK37-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32**
// CHECK37-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK37-NEXT: store i32* [[VLA]], i32** [[TMP26]], align 8
// CHECK37-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32** // CHECK37-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 3), align 8
// CHECK37-NEXT: store i32* [[VLA]], i32** [[TMP27]], align 8 // CHECK37-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK37-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK37-NEXT: store i8* null, i8** [[TMP27]], align 8
// CHECK37-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32** // CHECK37-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK37-NEXT: store i32* [[VLA]], i32** [[TMP29]], align 8 // CHECK37-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK37-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK37-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK37-NEXT: store i64 [[TMP7]], i64* [[TMP30]], align 8 // CHECK37-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK37-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK37-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK37-NEXT: store i8* null, i8** [[TMP31]], align 8 // CHECK37-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK37-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK37-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK37-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK37-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK37-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK37-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK37-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK37-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK37-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK37-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK37-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK37-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK37-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK37-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK37-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK37-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK37-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK37-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK37-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK37-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK37-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK37-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK37-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK37-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK37-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK37-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK37-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK37: omp_offload.failed: // CHECK37: omp_offload.failed:
// CHECK37-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK37-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK37-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK37-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK37: omp_offload.cont: // CHECK37: omp_offload.cont:
// CHECK37-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK37-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK37-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP41]]) // CHECK37-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP36]])
// CHECK37-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK37-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK37-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK37-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK37-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK37-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK37-NEXT: [[TMP43:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK37-NEXT: [[TMP38:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK37-NEXT: ret i32 [[TMP43]] // CHECK37-NEXT: ret i32 [[TMP38]]
// //
// //
// CHECK37-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 // CHECK37-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192
// CHECK37-SAME: (i64 [[ARGC:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK37-SAME: (i64 [[ARGC:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK37-NEXT: entry: // CHECK37-NEXT: entry:
// CHECK37-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8 // CHECK37-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8
// CHECK37-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK37-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK37-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK37-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 211 Lines • ▼ Show 20 Lines
// CHECK37-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK37-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK37-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 // CHECK37-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8
// CHECK37-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK37-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK37-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK37-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK37-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK37-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK37-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK37-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK37-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK37-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK37-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK37-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK37-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) // CHECK37-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1)
// CHECK37-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK37-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK37-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK37-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK37: omp_offload.failed: // CHECK37: omp_offload.failed:
// CHECK37-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK37-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK37-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK37-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK37: omp_offload.cont: // CHECK37: omp_offload.cont:
// CHECK37-NEXT: ret i32 0 // CHECK37-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 108 Lines • ▼ Show 20 Lines
// CHECK38-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK38-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK38-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK38-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK38-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK38-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK38-NEXT: [[ARGC_CASTED:%.*]] = alloca i64, align 8 // CHECK38-NEXT: [[ARGC_CASTED:%.*]] = alloca i64, align 8
// CHECK38-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK38-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK38-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK38-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK38-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 // CHECK38-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
// CHECK38-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 // CHECK38-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
// CHECK38-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 8
// CHECK38-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK38-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK38-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK38-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK38-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK38-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK38-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK38-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK38-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK38-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK38-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK38-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK38-NEXT: store i32 100, i32* [[N]], align 4 // CHECK38-NEXT: store i32 100, i32* [[N]], align 4
// CHECK38-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK38-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
Show All 12 Lines
// CHECK38-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK38-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK38-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK38-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK38-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK38-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK38-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK38-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK38-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK38-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK38-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK38-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK38-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64* // CHECK38-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i64*
// CHECK38-NEXT: store i64 [[TMP4]], i64* [[TMP11]], align 8 // CHECK38-NEXT: store i64 [[TMP4]], i64* [[TMP11]], align 8
// CHECK38-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK38-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK38-NEXT: store i64 4, i64* [[TMP12]], align 8 // CHECK38-NEXT: store i8* null, i8** [[TMP12]], align 8
// CHECK38-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK38-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK38-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK38-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK38-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK38-NEXT: store i64 [[TMP6]], i64* [[TMP14]], align 8
// CHECK38-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK38-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK38-NEXT: store i64 [[TMP6]], i64* [[TMP15]], align 8 // CHECK38-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i64*
// CHECK38-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK38-NEXT: store i64 [[TMP6]], i64* [[TMP16]], align 8
// CHECK38-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i64* // CHECK38-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK38-NEXT: store i64 [[TMP6]], i64* [[TMP17]], align 8 // CHECK38-NEXT: store i8* null, i8** [[TMP17]], align 8
// CHECK38-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK38-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK38-NEXT: store i64 4, i64* [[TMP18]], align 8 // CHECK38-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64*
// CHECK38-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK38-NEXT: store i64 [[TMP1]], i64* [[TMP19]], align 8
// CHECK38-NEXT: store i8* null, i8** [[TMP19]], align 8 // CHECK38-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK38-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK38-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK38-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64*
// CHECK38-NEXT: store i64 [[TMP1]], i64* [[TMP21]], align 8 // CHECK38-NEXT: store i64 [[TMP1]], i64* [[TMP21]], align 8
// CHECK38-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK38-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK38-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64* // CHECK38-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK38-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8 // CHECK38-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK38-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK38-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32**
// CHECK38-NEXT: store i64 8, i64* [[TMP24]], align 8 // CHECK38-NEXT: store i32* [[VLA]], i32** [[TMP24]], align 8
// CHECK38-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK38-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK38-NEXT: store i8* null, i8** [[TMP25]], align 8 // CHECK38-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32**
// CHECK38-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK38-NEXT: store i32* [[VLA]], i32** [[TMP26]], align 8
// CHECK38-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32** // CHECK38-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 3), align 8
// CHECK38-NEXT: store i32* [[VLA]], i32** [[TMP27]], align 8 // CHECK38-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK38-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK38-NEXT: store i8* null, i8** [[TMP27]], align 8
// CHECK38-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32** // CHECK38-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK38-NEXT: store i32* [[VLA]], i32** [[TMP29]], align 8 // CHECK38-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK38-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK38-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK38-NEXT: store i64 [[TMP7]], i64* [[TMP30]], align 8 // CHECK38-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK38-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK38-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK38-NEXT: store i8* null, i8** [[TMP31]], align 8 // CHECK38-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK38-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK38-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK38-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK38-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK38-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK38-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK38-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK38-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK38-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK38-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK38-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK38-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK38-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK38-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK38-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK38-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK38-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK38-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK38-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK38-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK38-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK38-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK38-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK38-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK38-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK38-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK38-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK38: omp_offload.failed: // CHECK38: omp_offload.failed:
// CHECK38-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK38-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK38-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK38-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK38: omp_offload.cont: // CHECK38: omp_offload.cont:
// CHECK38-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK38-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK38-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP41]]) // CHECK38-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP36]])
// CHECK38-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK38-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK38-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK38-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK38-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK38-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK38-NEXT: [[TMP43:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK38-NEXT: [[TMP38:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK38-NEXT: ret i32 [[TMP43]] // CHECK38-NEXT: ret i32 [[TMP38]]
// //
// //
// CHECK38-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 // CHECK38-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192
// CHECK38-SAME: (i64 [[ARGC:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK38-SAME: (i64 [[ARGC:%.*]], i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK38-NEXT: entry: // CHECK38-NEXT: entry:
// CHECK38-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8 // CHECK38-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8
// CHECK38-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK38-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK38-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK38-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 211 Lines • ▼ Show 20 Lines
// CHECK38-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK38-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK38-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 // CHECK38-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8
// CHECK38-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK38-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK38-NEXT: store i8* null, i8** [[TMP18]], align 8 // CHECK38-NEXT: store i8* null, i8** [[TMP18]], align 8
// CHECK38-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK38-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK38-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK38-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK38-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK38-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK38-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK38-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK38-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) // CHECK38-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1)
// CHECK38-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK38-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK38-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK38-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK38: omp_offload.failed: // CHECK38: omp_offload.failed:
// CHECK38-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK38-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK38-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK38-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK38: omp_offload.cont: // CHECK38: omp_offload.cont:
// CHECK38-NEXT: ret i32 0 // CHECK38-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 108 Lines • ▼ Show 20 Lines
// CHECK39-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK39-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK39-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK39-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK39-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK39-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK39-NEXT: [[ARGC_CASTED:%.*]] = alloca i32, align 4 // CHECK39-NEXT: [[ARGC_CASTED:%.*]] = alloca i32, align 4
// CHECK39-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK39-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK39-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK39-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK39-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK39-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK39-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 // CHECK39-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
// CHECK39-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 4
// CHECK39-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK39-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK39-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK39-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK39-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK39-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK39-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK39-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK39-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK39-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK39-NEXT: store i32 100, i32* [[N]], align 4 // CHECK39-NEXT: store i32 100, i32* [[N]], align 4
// CHECK39-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK39-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
Show All 10 Lines
// CHECK39-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK39-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK39-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64 // CHECK39-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64
// CHECK39-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK39-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK39-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK39-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK39-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK39-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK39-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK39-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK39-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32* // CHECK39-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32*
// CHECK39-NEXT: store i32 [[TMP3]], i32* [[TMP11]], align 4 // CHECK39-NEXT: store i32 [[TMP3]], i32* [[TMP11]], align 4
// CHECK39-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK39-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK39-NEXT: store i64 4, i64* [[TMP12]], align 4 // CHECK39-NEXT: store i8* null, i8** [[TMP12]], align 4
// CHECK39-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK39-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK39-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK39-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK39-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK39-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK39-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK39-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK39-NEXT: store i32 [[TMP5]], i32* [[TMP15]], align 4 // CHECK39-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK39-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK39-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK39-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK39-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK39-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK39-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK39-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK39-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK39-NEXT: store i64 4, i64* [[TMP18]], align 4 // CHECK39-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK39-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK39-NEXT: store i32 [[TMP0]], i32* [[TMP19]], align 4
// CHECK39-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK39-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK39-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK39-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32* // CHECK39-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK39-NEXT: store i32 [[TMP0]], i32* [[TMP21]], align 4 // CHECK39-NEXT: store i32 [[TMP0]], i32* [[TMP21]], align 4
// CHECK39-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK39-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK39-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK39-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK39-NEXT: store i32 [[TMP0]], i32* [[TMP23]], align 4 // CHECK39-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK39-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK39-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32**
// CHECK39-NEXT: store i64 4, i64* [[TMP24]], align 4 // CHECK39-NEXT: store i32* [[VLA]], i32** [[TMP24]], align 4
// CHECK39-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK39-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK39-NEXT: store i8* null, i8** [[TMP25]], align 4 // CHECK39-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32**
// CHECK39-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK39-NEXT: store i32* [[VLA]], i32** [[TMP26]], align 4
// CHECK39-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32** // CHECK39-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 3), align 4
// CHECK39-NEXT: store i32* [[VLA]], i32** [[TMP27]], align 4 // CHECK39-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK39-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK39-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK39-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32** // CHECK39-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK39-NEXT: store i32* [[VLA]], i32** [[TMP29]], align 4 // CHECK39-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK39-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK39-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK39-NEXT: store i64 [[TMP7]], i64* [[TMP30]], align 4 // CHECK39-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK39-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK39-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK39-NEXT: store i8* null, i8** [[TMP31]], align 4 // CHECK39-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK39-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK39-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK39-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK39-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK39-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK39-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK39-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK39-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK39-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK39-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK39-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK39-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK39-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK39-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK39-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK39-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK39-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK39-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK39-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK39-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK39-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK39-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK39-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK39-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK39-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK39-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK39-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK39: omp_offload.failed: // CHECK39: omp_offload.failed:
// CHECK39-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK39-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK39-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK39-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK39: omp_offload.cont: // CHECK39: omp_offload.cont:
// CHECK39-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK39-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK39-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP41]]) // CHECK39-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP36]])
// CHECK39-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK39-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK39-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK39-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK39-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK39-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK39-NEXT: [[TMP43:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK39-NEXT: [[TMP38:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK39-NEXT: ret i32 [[TMP43]] // CHECK39-NEXT: ret i32 [[TMP38]]
// //
// //
// CHECK39-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 // CHECK39-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192
// CHECK39-SAME: (i32 [[ARGC:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK39-SAME: (i32 [[ARGC:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK39-NEXT: entry: // CHECK39-NEXT: entry:
// CHECK39-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 // CHECK39-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK39-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK39-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK39-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK39-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 205 Lines • ▼ Show 20 Lines
// CHECK39-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK39-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK39-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 // CHECK39-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4
// CHECK39-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK39-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK39-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK39-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK39-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK39-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK39-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK39-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK39-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK39-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK39-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK39-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK39-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) // CHECK39-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1)
// CHECK39-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK39-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK39-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK39-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK39: omp_offload.failed: // CHECK39: omp_offload.failed:
// CHECK39-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK39-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK39-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK39-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK39: omp_offload.cont: // CHECK39: omp_offload.cont:
// CHECK39-NEXT: ret i32 0 // CHECK39-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 105 Lines • ▼ Show 20 Lines
// CHECK40-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK40-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK40-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK40-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK40-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK40-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK40-NEXT: [[ARGC_CASTED:%.*]] = alloca i32, align 4 // CHECK40-NEXT: [[ARGC_CASTED:%.*]] = alloca i32, align 4
// CHECK40-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK40-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK40-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK40-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK40-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 // CHECK40-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
// CHECK40-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 // CHECK40-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
// CHECK40-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 4
// CHECK40-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK40-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK40-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK40-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK40-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK40-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK40-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK40-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK40-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK40-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK40-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK40-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK40-NEXT: store i32 100, i32* [[N]], align 4 // CHECK40-NEXT: store i32 100, i32* [[N]], align 4
// CHECK40-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK40-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
Show All 10 Lines
// CHECK40-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK40-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK40-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64 // CHECK40-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64
// CHECK40-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK40-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK40-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK40-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK40-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK40-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK40-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK40-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK40-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32* // CHECK40-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to i32*
// CHECK40-NEXT: store i32 [[TMP3]], i32* [[TMP11]], align 4 // CHECK40-NEXT: store i32 [[TMP3]], i32* [[TMP11]], align 4
// CHECK40-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK40-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK40-NEXT: store i64 4, i64* [[TMP12]], align 4 // CHECK40-NEXT: store i8* null, i8** [[TMP12]], align 4
// CHECK40-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK40-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK40-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK40-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK40-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK40-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK40-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK40-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK40-NEXT: store i32 [[TMP5]], i32* [[TMP15]], align 4 // CHECK40-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to i32*
// CHECK40-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK40-NEXT: store i32 [[TMP5]], i32* [[TMP16]], align 4
// CHECK40-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32* // CHECK40-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK40-NEXT: store i32 [[TMP5]], i32* [[TMP17]], align 4 // CHECK40-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK40-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK40-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK40-NEXT: store i64 4, i64* [[TMP18]], align 4 // CHECK40-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK40-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK40-NEXT: store i32 [[TMP0]], i32* [[TMP19]], align 4
// CHECK40-NEXT: store i8* null, i8** [[TMP19]], align 4 // CHECK40-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK40-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK40-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32* // CHECK40-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32*
// CHECK40-NEXT: store i32 [[TMP0]], i32* [[TMP21]], align 4 // CHECK40-NEXT: store i32 [[TMP0]], i32* [[TMP21]], align 4
// CHECK40-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK40-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK40-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i32* // CHECK40-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK40-NEXT: store i32 [[TMP0]], i32* [[TMP23]], align 4 // CHECK40-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK40-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK40-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32**
// CHECK40-NEXT: store i64 4, i64* [[TMP24]], align 4 // CHECK40-NEXT: store i32* [[VLA]], i32** [[TMP24]], align 4
// CHECK40-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK40-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK40-NEXT: store i8* null, i8** [[TMP25]], align 4 // CHECK40-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32**
// CHECK40-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK40-NEXT: store i32* [[VLA]], i32** [[TMP26]], align 4
// CHECK40-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32** // CHECK40-NEXT: store i64 [[TMP7]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 3), align 4
// CHECK40-NEXT: store i32* [[VLA]], i32** [[TMP27]], align 4 // CHECK40-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK40-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK40-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK40-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32** // CHECK40-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK40-NEXT: store i32* [[VLA]], i32** [[TMP29]], align 4 // CHECK40-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK40-NEXT: [[TMP30:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK40-NEXT: [[TMP30:%.*]] = load i32, i32* [[N]], align 4
// CHECK40-NEXT: store i64 [[TMP7]], i64* [[TMP30]], align 4 // CHECK40-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK40-NEXT: [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK40-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK40-NEXT: store i8* null, i8** [[TMP31]], align 4 // CHECK40-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK40-NEXT: [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK40-NEXT: [[TMP33:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK40-NEXT: [[TMP34:%.*]] = getelementptr inbounds [4 x i64], [4 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK40-NEXT: [[TMP35:%.*]] = load i32, i32* [[N]], align 4
// CHECK40-NEXT: store i32 [[TMP35]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK40-NEXT: [[TMP36:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK40-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK40-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK40-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK40-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK40-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK40-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK40-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK40-NEXT: [[TMP37:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK40-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK40-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP37]], 1 // CHECK40-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK40-NEXT: [[TMP38:%.*]] = zext i32 [[ADD]] to i64 // CHECK40-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64
// CHECK40-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP38]]) // CHECK40-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP33]])
// CHECK40-NEXT: [[TMP39:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 4, i8** [[TMP32]], i8** [[TMP33]], i64* [[TMP34]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK40-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, i32 4, i8** [[TMP28]], i8** [[TMP29]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK40-NEXT: [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0 // CHECK40-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK40-NEXT: br i1 [[TMP40]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK40-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK40: omp_offload.failed: // CHECK40: omp_offload.failed:
// CHECK40-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK40-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK40-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK40-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK40: omp_offload.cont: // CHECK40: omp_offload.cont:
// CHECK40-NEXT: [[TMP41:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK40-NEXT: [[TMP36:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK40-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP41]]) // CHECK40-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP36]])
// CHECK40-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK40-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK40-NEXT: [[TMP42:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK40-NEXT: [[TMP37:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK40-NEXT: call void @llvm.stackrestore(i8* [[TMP42]]) // CHECK40-NEXT: call void @llvm.stackrestore(i8* [[TMP37]])
// CHECK40-NEXT: [[TMP43:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK40-NEXT: [[TMP38:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK40-NEXT: ret i32 [[TMP43]] // CHECK40-NEXT: ret i32 [[TMP38]]
// //
// //
// CHECK40-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 // CHECK40-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192
// CHECK40-SAME: (i32 [[ARGC:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK40-SAME: (i32 [[ARGC:%.*]], i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK40-NEXT: entry: // CHECK40-NEXT: entry:
// CHECK40-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 // CHECK40-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK40-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK40-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK40-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK40-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 205 Lines • ▼ Show 20 Lines
// CHECK40-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** // CHECK40-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]**
// CHECK40-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 // CHECK40-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4
// CHECK40-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK40-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK40-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK40-NEXT: store i8* null, i8** [[TMP18]], align 4
// CHECK40-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK40-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK40-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK40-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK40-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 // CHECK40-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4
// CHECK40-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK40-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK40-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1) // CHECK40-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 1)
// CHECK40-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 // CHECK40-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0
// CHECK40-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK40-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK40: omp_offload.failed: // CHECK40: omp_offload.failed:
// CHECK40-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] // CHECK40-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]]
// CHECK40-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK40-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK40: omp_offload.cont: // CHECK40: omp_offload.cont:
// CHECK40-NEXT: ret i32 0 // CHECK40-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 1,288 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_distribute_simd_collapse_codegen.cpp

Show First 20 Lines • Show All 188 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: br label [[COND_END]] // CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end: // CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond: // CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body: // CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 456 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 456
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 456 // CHECK1-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 456
// CHECK1-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456 // CHECK1-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]] // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
// CHECK1-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK1-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK1-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK1-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i64 0, i64 [[IDXPROM]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64 // CHECK1-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64
// CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]] // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
// CHECK1-NEXT: store i32 0, i32* [[ARRAYIDX8]], align 4 // CHECK1-NEXT: store i32 0, i32* [[ARRAYIDX8]], align 4, !llvm.access.group !4
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue: // CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc: // CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK1-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK1: omp.inner.for.end: // CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit: // CHECK1: omp.loop.exit:
// CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then: // CHECK1: .omp.final.then:
▲ Show 20 Lines • Show All 102 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: br label [[COND_END]] // CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end: // CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond: // CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body: // CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 456 // CHECK2-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 456
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 456 // CHECK2-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 456
// CHECK2-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456 // CHECK2-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK2-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]] // CHECK2-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
// CHECK2-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK2-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK2-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK2-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 // CHECK2-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 // CHECK2-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i64 0, i64 [[IDXPROM]] // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i64 0, i64 [[IDXPROM]]
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4 // CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64 // CHECK2-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64
// CHECK2-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]] // CHECK2-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
// CHECK2-NEXT: store i32 0, i32* [[ARRAYIDX8]], align 4 // CHECK2-NEXT: store i32 0, i32* [[ARRAYIDX8]], align 4, !llvm.access.group !4
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue: // CHECK2: omp.body.continue:
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc: // CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK2-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK2-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK2-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK2: omp.inner.for.end: // CHECK2: omp.inner.for.end:
// CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2: omp.loop.exit: // CHECK2: omp.loop.exit:
// CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK2-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK2-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK2-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK2: .omp.final.then: // CHECK2: .omp.final.then:
▲ Show 20 Lines • Show All 102 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: br label [[COND_END]] // CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end: // CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond: // CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body: // CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 456 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 456
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 456 // CHECK3-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 456
// CHECK3-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456 // CHECK3-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]] // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
// CHECK3-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK3-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK3-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK3-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i32 0, i32 [[TMP11]] // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i32 0, i32 [[TMP11]]
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP12]] // CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP12]]
// CHECK3-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4 // CHECK3-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4, !llvm.access.group !5
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3: omp.body.continue: // CHECK3: omp.body.continue:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3: omp.inner.for.inc: // CHECK3: omp.inner.for.inc:
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK3-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK3: omp.inner.for.end: // CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit: // CHECK3: omp.loop.exit:
// CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK3: .omp.final.then: // CHECK3: .omp.final.then:
▲ Show 20 Lines • Show All 102 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: br label [[COND_END]] // CHECK4-NEXT: br label [[COND_END]]
// CHECK4: cond.end: // CHECK4: cond.end:
// CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK4-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK4-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK4-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK4: omp.inner.for.cond: // CHECK4: omp.inner.for.cond:
// CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK4-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK4-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK4: omp.inner.for.body: // CHECK4: omp.inner.for.body:
// CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 456 // CHECK4-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 456
// CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK4-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 456 // CHECK4-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 456
// CHECK4-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456 // CHECK4-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK4-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]] // CHECK4-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
// CHECK4-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK4-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK4-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK4-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK4-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK4-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 // CHECK4-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0
// CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i32 0, i32 [[TMP11]] // CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], [123 x [456 x i32]]* [[A]], i32 0, i32 [[TMP11]]
// CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP12]] // CHECK4-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], [456 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP12]]
// CHECK4-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4 // CHECK4-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4, !llvm.access.group !5
// CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK4-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK4: omp.body.continue: // CHECK4: omp.body.continue:
// CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK4: omp.inner.for.inc: // CHECK4: omp.inner.for.inc:
// CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK4-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK4-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK4-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !5
// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK4: omp.inner.for.end: // CHECK4: omp.inner.for.end:
// CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK4-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK4: omp.loop.exit: // CHECK4: omp.loop.exit:
// CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK4-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK4-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK4-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK4-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK4-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK4: .omp.final.then: // CHECK4: .omp.final.then:
▲ Show 20 Lines • Show All 302 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK9-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK9-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK9-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK9-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK9-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK9-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK9-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK9-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK9-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK9-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK9-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK9-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK9-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK9-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK9-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP54]] // CHECK9-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 94 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: br label [[COND_END]] // CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end: // CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond: // CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK9-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !5
// CHECK9-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0 // CHECK9-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0
// CHECK9-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1 // CHECK9-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK9-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]] // CHECK9-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK9-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64 // CHECK9-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK9-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]] // CHECK9-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]]
// CHECK9-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1 // CHECK9-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK9-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32 // CHECK9-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK9-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4 // CHECK9-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4, !llvm.access.group !5
// CHECK9-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !5
// CHECK9-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0 // CHECK9-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK9-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1 // CHECK9-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK9-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]] // CHECK9-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK9-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64 // CHECK9-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK9-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]] // CHECK9-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]]
// CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !5
// CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]] // CHECK9-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK9-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64 // CHECK9-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]] // CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK9-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]] // CHECK9-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]]
// CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1 // CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK9-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]] // CHECK9-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK9-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32 // CHECK9-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK9-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4 // CHECK9-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4, !llvm.access.group !5
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I11]], align 4 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[I11]], align 4, !llvm.access.group !5
// CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP27]] to i64 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP27]] to i64
// CHECK9-NEXT: [[TMP28:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]] // CHECK9-NEXT: [[TMP28:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]]
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i64 [[TMP28]] // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i64 [[TMP28]]
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[J12]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[J12]], align 4, !llvm.access.group !5
// CHECK9-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP29]] to i64 // CHECK9-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP29]] to i64
// CHECK9-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 [[IDXPROM36]] // CHECK9-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 [[IDXPROM36]]
// CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX37]], align 4 // CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX37]], align 4, !llvm.access.group !5
// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9: omp.body.continue: // CHECK9: omp.body.continue:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP30:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK9-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP30]], 1 // CHECK9-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP30]], 1
// CHECK9-NEXT: store i64 [[ADD38]], i64* [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[ADD38]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP32]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP32]])
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK9-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
Show All 35 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: br label [[COND_END]] // CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end: // CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond: // CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !11
// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 2 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 2
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !11
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK9-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 2 // CHECK9-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 2
// CHECK9-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2 // CHECK9-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]] // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
// CHECK9-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK9-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK9-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK9-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !11
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11
// CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i64 0, i64 [[IDXPROM]] // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i64 0, i64 [[IDXPROM]]
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !11
// CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64
// CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]] // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
// CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX8]], align 4 // CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX8]], align 4, !llvm.access.group !11
// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9: omp.body.continue: // CHECK9: omp.body.continue:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc: // CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK9-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit: // CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
Show All 22 Lines
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
Show All 20 Lines
// CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK10-NEXT: [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4 // CHECK10-NEXT: [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64*
// CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i64 [[TMP7]], i64* [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP16]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP16]], align 8
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64*
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP18]], align 8
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i64* // CHECK10-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP19]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64*
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP20]], align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i64* // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP9]], i64* [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 4, i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to i64*
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP23]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64* // CHECK10-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP25]], align 8
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64* // CHECK10-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP27]], align 8 // CHECK10-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK10-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK10-NEXT: store i64 8, i64* [[TMP28]], align 8 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP28]], align 8
// CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK10-NEXT: store i8* null, i8** [[TMP29]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK10-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP30]], align 8
// CHECK10-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK10-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP31]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i32**
// CHECK10-NEXT: store i64 [[TMP3]], i64* [[TMP33]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP33]], align 8
// CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK10-NEXT: store i64 8, i64* [[TMP34]], align 8 // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i32**
// CHECK10-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP35]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP35]], align 8 // CHECK10-NEXT: store i64 [[TMP11]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 8
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32** // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP37]], align 8 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32** // CHECK10-NEXT: [[TMP39:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP39]], align 8 // CHECK10-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK10-NEXT: [[TMP40:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i64 [[TMP11]], i64* [[TMP40]], align 8 // CHECK10-NEXT: store i32 [[TMP40]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK10-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP45:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[M]], align 4
// CHECK10-NEXT: store i32 [[TMP46]], i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP47]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64 // CHECK10-NEXT: [[CONV5:%.*]] = sext i32 [[DIV]] to i64
// CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 // CHECK10-NEXT: [[TMP42:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4
// CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP48]], 0 // CHECK10-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP42]], 0
// CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK10-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK10-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]]
// CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK10-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: store i64 [[SUB9]], i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[TMP49:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8 // CHECK10-NEXT: [[TMP43:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_4]], align 8
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP49]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP43]], 1
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK10-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP42]], i8** [[TMP43]], i64* [[TMP44]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP37]], i8** [[TMP38]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0 // CHECK10-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK10-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP45]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP52:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: [[TMP46:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP52]]) // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 signext [[TMP46]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP53]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP47]])
// CHECK10-NEXT: [[TMP54:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP48:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP54]] // CHECK10-NEXT: ret i32 [[TMP48]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[M:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 94 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: br label [[COND_END]] // CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end: // CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ] // CHECK10-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 // CHECK10-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK10-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK10-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond: // CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 // CHECK10-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK10-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK10-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK10-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !5
// CHECK10-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0 // CHECK10-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0
// CHECK10-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1 // CHECK10-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK10-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]] // CHECK10-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK10-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64 // CHECK10-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK10-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]] // CHECK10-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]]
// CHECK10-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1 // CHECK10-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK10-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32 // CHECK10-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK10-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4 // CHECK10-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4, !llvm.access.group !5
// CHECK10-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !5
// CHECK10-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0 // CHECK10-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK10-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1 // CHECK10-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK10-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]] // CHECK10-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK10-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64 // CHECK10-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK10-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]] // CHECK10-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]]
// CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK10-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !5
// CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0 // CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK10-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1 // CHECK10-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK10-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]] // CHECK10-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK10-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64 // CHECK10-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK10-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]] // CHECK10-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK10-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]] // CHECK10-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]]
// CHECK10-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1 // CHECK10-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK10-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]] // CHECK10-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK10-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32 // CHECK10-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK10-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4 // CHECK10-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4, !llvm.access.group !5
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[I11]], align 4 // CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[I11]], align 4, !llvm.access.group !5
// CHECK10-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP27]] to i64 // CHECK10-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP27]] to i64
// CHECK10-NEXT: [[TMP28:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]] // CHECK10-NEXT: [[TMP28:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]]
// CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i64 [[TMP28]] // CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i64 [[TMP28]]
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[J12]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[J12]], align 4, !llvm.access.group !5
// CHECK10-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP29]] to i64 // CHECK10-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP29]] to i64
// CHECK10-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 [[IDXPROM36]] // CHECK10-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 [[IDXPROM36]]
// CHECK10-NEXT: store i32 0, i32* [[ARRAYIDX37]], align 4 // CHECK10-NEXT: store i32 0, i32* [[ARRAYIDX37]], align 4, !llvm.access.group !5
// CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10: omp.body.continue: // CHECK10: omp.body.continue:
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP30:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK10-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP30]], 1 // CHECK10-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP30]], 1
// CHECK10-NEXT: store i64 [[ADD38]], i64* [[DOTOMP_IV]], align 8 // CHECK10-NEXT: store i64 [[ADD38]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !5
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK10-NEXT: [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4 // CHECK10-NEXT: [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP32]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP32]])
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK10-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
Show All 35 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: br label [[COND_END]] // CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end: // CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond: // CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !11
// CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body: // CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 2 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 2
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !11
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK10-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 2 // CHECK10-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 2
// CHECK10-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2 // CHECK10-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]] // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
// CHECK10-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK10-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK10-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK10-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK10-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK10-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !11
// CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 // CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !11
// CHECK10-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 // CHECK10-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i64 0, i64 [[IDXPROM]] // CHECK10-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i64 0, i64 [[IDXPROM]]
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4 // CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !11
// CHECK10-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64 // CHECK10-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64
// CHECK10-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]] // CHECK10-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
// CHECK10-NEXT: store i32 0, i32* [[ARRAYIDX8]], align 4 // CHECK10-NEXT: store i32 0, i32* [[ARRAYIDX8]], align 4, !llvm.access.group !11
// CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10: omp.body.continue: // CHECK10: omp.body.continue:
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc: // CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK10-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK10-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK10-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4 // CHECK10-NEXT: store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !11
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]] // CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK10: omp.inner.for.end: // CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit: // CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK10-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK10: .omp.final.then: // CHECK10: .omp.final.then:
Show All 22 Lines
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK11-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK11-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK11-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK11-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK11-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK11-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK11-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK11-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK11-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK11-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK11-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK11-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK11-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK11-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK11-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK11-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK11-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK11-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP53]] // CHECK11-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 92 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: br label [[COND_END]] // CHECK11-NEXT: br label [[COND_END]]
// CHECK11: cond.end: // CHECK11: cond.end:
// CHECK11-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ] // CHECK11-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 // CHECK11-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK11-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 // CHECK11-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK11-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 // CHECK11-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK11-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK11-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !6
// CHECK11-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0 // CHECK11-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0
// CHECK11-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1 // CHECK11-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK11-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]] // CHECK11-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK11-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64 // CHECK11-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK11-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]] // CHECK11-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]]
// CHECK11-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1 // CHECK11-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK11-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32 // CHECK11-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK11-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4 // CHECK11-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4, !llvm.access.group !6
// CHECK11-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !6
// CHECK11-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0 // CHECK11-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK11-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1 // CHECK11-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK11-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]] // CHECK11-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK11-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64 // CHECK11-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK11-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]] // CHECK11-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]]
// CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !6
// CHECK11-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0 // CHECK11-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK11-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1 // CHECK11-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK11-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]] // CHECK11-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK11-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64 // CHECK11-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK11-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]] // CHECK11-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK11-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]] // CHECK11-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]]
// CHECK11-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1 // CHECK11-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK11-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]] // CHECK11-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK11-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32 // CHECK11-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK11-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4 // CHECK11-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4, !llvm.access.group !6
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I11]], align 4 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[I11]], align 4, !llvm.access.group !6
// CHECK11-NEXT: [[TMP28:%.*]] = mul nsw i32 [[TMP27]], [[TMP3]] // CHECK11-NEXT: [[TMP28:%.*]] = mul nsw i32 [[TMP27]], [[TMP3]]
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i32 [[TMP28]] // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i32 [[TMP28]]
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[J12]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[J12]], align 4, !llvm.access.group !6
// CHECK11-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i32 [[TMP29]] // CHECK11-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i32 [[TMP29]]
// CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX36]], align 4 // CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX36]], align 4, !llvm.access.group !6
// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11: omp.body.continue: // CHECK11: omp.body.continue:
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP30:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: [[TMP30:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK11-NEXT: [[ADD37:%.*]] = add nsw i64 [[TMP30]], 1 // CHECK11-NEXT: [[ADD37:%.*]] = add nsw i64 [[TMP30]], 1
// CHECK11-NEXT: store i64 [[ADD37]], i64* [[DOTOMP_IV]], align 8 // CHECK11-NEXT: store i64 [[ADD37]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK11-NEXT: [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP32]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP32]])
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK11-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
Show All 35 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: br label [[COND_END]] // CHECK11-NEXT: br label [[COND_END]]
// CHECK11: cond.end: // CHECK11: cond.end:
// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK11-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11: omp.inner.for.cond: // CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !12
// CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body: // CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 2 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 2
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !12
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK11-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 2 // CHECK11-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 2
// CHECK11-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2 // CHECK11-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]] // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
// CHECK11-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK11-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK11-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK11-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !12
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !12
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i32 0, i32 [[TMP11]] // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i32 0, i32 [[TMP11]]
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !12
// CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP12]] // CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP12]]
// CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4 // CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4, !llvm.access.group !12
// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11: omp.body.continue: // CHECK11: omp.body.continue:
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc: // CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK11-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]] // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK11: omp.inner.for.end: // CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit: // CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK11-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK11-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK11-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK11: .omp.final.then: // CHECK11: .omp.final.then:
Show All 22 Lines
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
Show All 17 Lines
// CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4 // CHECK12-NEXT: [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64 // CHECK12-NEXT: [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32* // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32*
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP17]], align 4
// CHECK12-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP18]], align 4 // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32*
// CHECK12-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP7]], i32* [[TMP20]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP21]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to i32*
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP22]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32* // CHECK12-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK12-NEXT: store i8* null, i8** [[TMP25]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK12-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
// CHECK12-NEXT: store i64 4, i64* [[TMP27]], align 4 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP27]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK12-NEXT: store i8* null, i8** [[TMP28]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32*
// CHECK12-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP29]], align 4
// CHECK12-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i32* // CHECK12-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP30]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP30]], align 4
// CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK12-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32* // CHECK12-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i32**
// CHECK12-NEXT: store i32 [[TMP1]], i32* [[TMP32]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP32]], align 4
// CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK12-NEXT: store i64 4, i64* [[TMP33]], align 4 // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32**
// CHECK12-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP34]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP34]], align 4 // CHECK12-NEXT: store i64 [[TMP10]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 4), align 4
// CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32** // CHECK12-NEXT: store i8* null, i8** [[TMP35]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP36]], align 4 // CHECK12-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32** // CHECK12-NEXT: [[TMP38:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP38]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK12-NEXT: [[TMP39:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i64 [[TMP10]], i64* [[TMP39]], align 4 // CHECK12-NEXT: store i32 [[TMP39]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK12-NEXT: [[TMP40:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP40]], align 4 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP40]], 0
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [5 x i64], [5 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP44:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP44]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[M]], align 4
// CHECK12-NEXT: store i32 [[TMP45]], i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[TMP46:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP46]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64 // CHECK12-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4 // CHECK12-NEXT: [[TMP41:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_2]], align 4
// CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP47]], 0 // CHECK12-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP41]], 0
// CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1 // CHECK12-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64 // CHECK12-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]] // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK12-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: store i64 [[SUB7]], i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[TMP48:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8 // CHECK12-NEXT: [[TMP42:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR_3]], align 8
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP48]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP42]], 1
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[ADD]])
// CHECK12-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP41]], i8** [[TMP42]], i64* [[TMP43]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83.region_id, i32 5, i8** [[TMP36]], i8** [[TMP37]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 // CHECK12-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
// CHECK12-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: [[TMP45:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP51]]) // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10ELi2EEiT_(i32 [[TMP45]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP46:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP52]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP46]])
// CHECK12-NEXT: [[TMP53:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP53]] // CHECK12-NEXT: ret i32 [[TMP47]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l83
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[M:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 92 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: br label [[COND_END]] // CHECK12-NEXT: br label [[COND_END]]
// CHECK12: cond.end: // CHECK12: cond.end:
// CHECK12-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ] // CHECK12-NEXT: [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK12-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 // CHECK12-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK12-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 // CHECK12-NEXT: [[TMP18:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK12-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: store i64 [[TMP18]], i64* [[DOTOMP_IV]], align 8
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP19:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 // CHECK12-NEXT: [[TMP20:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK12-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK12-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP21:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !6
// CHECK12-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0 // CHECK12-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0
// CHECK12-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1 // CHECK12-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK12-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]] // CHECK12-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK12-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64 // CHECK12-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK12-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]] // CHECK12-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]]
// CHECK12-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1 // CHECK12-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK12-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32 // CHECK12-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK12-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4 // CHECK12-NEXT: store i32 [[CONV21]], i32* [[I11]], align 4, !llvm.access.group !6
// CHECK12-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP23:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP24:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !6
// CHECK12-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0 // CHECK12-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK12-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1 // CHECK12-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK12-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]] // CHECK12-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK12-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64 // CHECK12-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK12-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]] // CHECK12-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]]
// CHECK12-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4 // CHECK12-NEXT: [[TMP26:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !6
// CHECK12-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0 // CHECK12-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK12-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1 // CHECK12-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK12-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]] // CHECK12-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK12-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64 // CHECK12-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK12-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]] // CHECK12-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK12-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]] // CHECK12-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]]
// CHECK12-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1 // CHECK12-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK12-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]] // CHECK12-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK12-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32 // CHECK12-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK12-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4 // CHECK12-NEXT: store i32 [[CONV35]], i32* [[J12]], align 4, !llvm.access.group !6
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[I11]], align 4 // CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[I11]], align 4, !llvm.access.group !6
// CHECK12-NEXT: [[TMP28:%.*]] = mul nsw i32 [[TMP27]], [[TMP3]] // CHECK12-NEXT: [[TMP28:%.*]] = mul nsw i32 [[TMP27]], [[TMP3]]
// CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i32 [[TMP28]] // CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i32 [[TMP28]]
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[J12]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[J12]], align 4, !llvm.access.group !6
// CHECK12-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i32 [[TMP29]] // CHECK12-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i32 [[TMP29]]
// CHECK12-NEXT: store i32 0, i32* [[ARRAYIDX36]], align 4 // CHECK12-NEXT: store i32 0, i32* [[ARRAYIDX36]], align 4, !llvm.access.group !6
// CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK12: omp.body.continue: // CHECK12: omp.body.continue:
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP30:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: [[TMP30:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK12-NEXT: [[ADD37:%.*]] = add nsw i64 [[TMP30]], 1 // CHECK12-NEXT: [[ADD37:%.*]] = add nsw i64 [[TMP30]], 1
// CHECK12-NEXT: store i64 [[ADD37]], i64* [[DOTOMP_IV]], align 8 // CHECK12-NEXT: store i64 [[ADD37]], i64* [[DOTOMP_IV]], align 8, !llvm.access.group !6
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK12-NEXT: [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK12-NEXT: [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4 // CHECK12-NEXT: [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP32]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP32]])
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK12-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
Show All 35 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x [2 x i32]]**
// CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x [2 x i32]]* [[A]], [10 x [2 x i32]]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 20)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l69([10 x [2 x i32]]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: br label [[COND_END]] // CHECK12-NEXT: br label [[COND_END]]
// CHECK12: cond.end: // CHECK12: cond.end:
// CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK12-NEXT: [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 // CHECK12-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 // CHECK12-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK12-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK12: omp.inner.for.cond: // CHECK12: omp.inner.for.cond:
// CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 // CHECK12-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4, !llvm.access.group !12
// CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK12-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK12-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12: omp.inner.for.body: // CHECK12: omp.inner.for.body:
// CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 2 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP8]], 2
// CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4 // CHECK12-NEXT: store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !12
// CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK12-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 2 // CHECK12-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP10]], 2
// CHECK12-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2 // CHECK12-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]] // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP9]], [[MUL4]]
// CHECK12-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK12-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK12-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK12-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK12-NEXT: store i32 [[ADD6]], i32* [[J]], align 4 // CHECK12-NEXT: store i32 [[ADD6]], i32* [[J]], align 4, !llvm.access.group !12
// CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 // CHECK12-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4, !llvm.access.group !12
// CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i32 0, i32 [[TMP11]] // CHECK12-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], [10 x [2 x i32]]* [[TMP0]], i32 0, i32 [[TMP11]]
// CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = load i32, i32* [[J]], align 4, !llvm.access.group !12
// CHECK12-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP12]] // CHECK12-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[ARRAYIDX]], i32 0, i32 [[TMP12]]
// CHECK12-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4 // CHECK12-NEXT: store i32 0, i32* [[ARRAYIDX7]], align 4, !llvm.access.group !12
// CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK12-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK12: omp.body.continue: // CHECK12: omp.body.continue:
// CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12: omp.inner.for.inc: // CHECK12: omp.inner.for.inc:
// CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK12-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK12-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK12-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4 // CHECK12-NEXT: store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !12
// CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]] // CHECK12-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK12: omp.inner.for.end: // CHECK12: omp.inner.for.end:
// CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK12-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK12: omp.loop.exit: // CHECK12: omp.loop.exit:
// CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) // CHECK12-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK12-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK12-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK12-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK12: .omp.final.then: // CHECK12: .omp.final.then:
▲ Show 20 Lines • Show All 813 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_distribute_simd_dist_schedule_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 2,098 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[N_CASTED17:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [3 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [3 x i8*], align 8 // CHECK9-NEXT: [[_TMP22:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES23:%.*]] = alloca [3 x i64], align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, i32* [[N]], align 4 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK9-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK9-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK9-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK9-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK9-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK9-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK9-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK9-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK9-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK9-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK9-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK9-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK9-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK9-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK9-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK9-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK9-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK9-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK9-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 8
// CHECK9-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK9-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK9-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK9-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK9-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK9-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK9-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK9-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK9-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK9-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK9-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP52]])
// CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK9-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK9-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK9: omp_offload.failed15:
// CHECK9-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK9-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK9: omp_offload.cont16:
// CHECK9-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: [[TMP55:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK9-NEXT: [[CONV18:%.*]] = bitcast i64* [[N_CASTED17]] to i32*
// CHECK9-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK9-NEXT: store i32 [[TMP55]], i32* [[CONV18]], align 4
// CHECK9-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK9-NEXT: [[TMP56:%.*]] = load i64, i64* [[N_CASTED17]], align 8
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP60]]) // CHECK9-NEXT: [[TMP57:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP58:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK9-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK9-NEXT: [[TMP59:%.*]] = bitcast i8** [[TMP58]] to i64*
// CHECK9-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK9-NEXT: store i64 [[TMP56]], i64* [[TMP59]], align 8
// CHECK9: omp_offload.failed16: // CHECK9-NEXT: [[TMP60:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK9-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i64*
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK9-NEXT: store i64 [[TMP56]], i64* [[TMP61]], align 8
// CHECK9: omp_offload.cont17: // CHECK9-NEXT: [[TMP62:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 0
// CHECK9-NEXT: [[TMP63:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: store i8* null, i8** [[TMP62]], align 8
// CHECK9-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32* // CHECK9-NEXT: [[TMP63:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1
// CHECK9-NEXT: store i32 [[TMP63]], i32* [[CONV19]], align 4 // CHECK9-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK9-NEXT: [[TMP64:%.*]] = load i64, i64* [[N_CASTED18]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP64]], align 8
// CHECK9-NEXT: [[TMP65:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK9-NEXT: [[TMP65:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1
// CHECK9-NEXT: [[TMP66:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK9-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i64*
// CHECK9-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64* // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP66]], align 8
// CHECK9-NEXT: store i64 [[TMP64]], i64* [[TMP67]], align 8 // CHECK9-NEXT: [[TMP67:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 1
// CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK9-NEXT: store i8* null, i8** [[TMP67]], align 8
// CHECK9-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64* // CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP64]], i64* [[TMP69]], align 8 // CHECK9-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32**
// CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 0 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP69]], align 8
// CHECK9-NEXT: store i64 4, i64* [[TMP70]], align 8 // CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2
// CHECK9-NEXT: [[TMP71:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK9-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32**
// CHECK9-NEXT: store i8* null, i8** [[TMP71]], align 8 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP71]], align 8
// CHECK9-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK9-NEXT: store i64 [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 2), align 8
// CHECK9-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i64* // CHECK9-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 2
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP73]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP72]], align 8
// CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK9-NEXT: [[TMP73:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK9-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64* // CHECK9-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP75]], align 8 // CHECK9-NEXT: [[TMP75:%.*]] = load i32, i32* [[N]], align 4
// CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 1 // CHECK9-NEXT: store i32 [[TMP75]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK9-NEXT: store i64 8, i64* [[TMP76]], align 8 // CHECK9-NEXT: [[TMP76:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK9-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP76]], 0
// CHECK9-NEXT: store i8* null, i8** [[TMP77]], align 8 // CHECK9-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK9-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32** // CHECK9-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP79]], align 8 // CHECK9-NEXT: [[TMP77:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK9-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP77]], 1
// CHECK9-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32** // CHECK9-NEXT: [[TMP78:%.*]] = zext i32 [[ADD28]] to i64
// CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP81]], align 8 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP78]])
// CHECK9-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 2 // CHECK9-NEXT: [[TMP79:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP73]], i8** [[TMP74]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: store i64 [[TMP65]], i64* [[TMP82]], align 8 // CHECK9-NEXT: [[TMP80:%.*]] = icmp ne i32 [[TMP79]], 0
// CHECK9-NEXT: [[TMP83:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK9-NEXT: br i1 [[TMP80]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK9-NEXT: store i8* null, i8** [[TMP83]], align 8 // CHECK9: omp_offload.failed29:
// CHECK9-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i64 [[TMP56]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: [[TMP85:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK9-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 0 // CHECK9: omp_offload.cont30:
// CHECK9-NEXT: [[TMP87:%.*]] = load i32, i32* [[N]], align 4 // CHECK9-NEXT: [[TMP81:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP87]], i32* [[DOTCAPTURE_EXPR_25]], align 4 // CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP81]])
// CHECK9-NEXT: [[TMP88:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP88]], 0
// CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK9-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[TMP89:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK9-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP89]], 1
// CHECK9-NEXT: [[TMP90:%.*]] = zext i32 [[ADD30]] to i64
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP90]])
// CHECK9-NEXT: [[TMP91:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP84]], i8** [[TMP85]], i64* [[TMP86]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP92:%.*]] = icmp ne i32 [[TMP91]], 0
// CHECK9-NEXT: br i1 [[TMP92]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK9: omp_offload.failed31:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i64 [[TMP64]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK9: omp_offload.cont32:
// CHECK9-NEXT: [[TMP93:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP93]])
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP94:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP82:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP94]]) // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP82]])
// CHECK9-NEXT: [[TMP95:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP83:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP95]] // CHECK9-NEXT: ret i32 [[TMP83]]
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 252 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK9-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 140 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK9-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed: // CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont: // CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK9-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK9: omp_offload.failed5: // CHECK9: omp_offload.failed5:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK9: omp_offload.cont6: // CHECK9: omp_offload.cont6:
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK9-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK9-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8 // CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK9-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK9-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK9-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK9-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK9-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK9: omp_offload.failed11: // CHECK9: omp_offload.failed11:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK9: omp_offload.cont12: // CHECK9: omp_offload.cont12:
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 94 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 // CHECK10-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK10-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES8:%.*]] = alloca [3 x i64], align 8 // CHECK10-NEXT: [[_TMP8:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP9:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_11:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[N_CASTED17:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [3 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [3 x i8*], align 8 // CHECK10-NEXT: [[_TMP22:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOFFLOAD_SIZES23:%.*]] = alloca [3 x i64], align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK10-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 // CHECK10-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK10-NEXT: store i32 100, i32* [[N]], align 4 // CHECK10-NEXT: store i32 100, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK10-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() // CHECK10-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave()
// CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 // CHECK10-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* // CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* // CHECK10-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP7]], align 8
// CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64* // CHECK10-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i64*
// CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8 // CHECK10-NEXT: store i64 [[TMP4]], i64* [[TMP9]], align 8
// CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK10-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP10]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8 // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64*
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 // CHECK10-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i64*
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP14]], align 8
// CHECK10-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP15]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK10-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: store i64 8, i64* [[TMP16]], align 8 // CHECK10-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK10-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP17]], align 8 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK10-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP20]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK10-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 8 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK10-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP23]], align 8 // CHECK10-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK10-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK10-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK10-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK10-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK10-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK10-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK10-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK10-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK10-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK10-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32* // CHECK10-NEXT: [[CONV4:%.*]] = bitcast i64* [[N_CASTED3]] to i32*
// CHECK10-NEXT: store i32 [[TMP33]], i32* [[CONV4]], align 4 // CHECK10-NEXT: store i32 [[TMP29]], i32* [[CONV4]], align 4
// CHECK10-NEXT: [[TMP34:%.*]] = load i64, i64* [[N_CASTED3]], align 8 // CHECK10-NEXT: [[TMP30:%.*]] = load i64, i64* [[N_CASTED3]], align 8
// CHECK10-NEXT: [[TMP35:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP31:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i64* // CHECK10-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP37]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP33]], align 8
// CHECK10-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* // CHECK10-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to i64*
// CHECK10-NEXT: store i64 [[TMP34]], i64* [[TMP39]], align 8 // CHECK10-NEXT: store i64 [[TMP30]], i64* [[TMP35]], align 8
// CHECK10-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK10-NEXT: store i64 4, i64* [[TMP40]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 // CHECK10-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK10-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP38]], align 8
// CHECK10-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK10-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP40]], align 8
// CHECK10-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 // CHECK10-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK10-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP43]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP43]], align 8
// CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 // CHECK10-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64* // CHECK10-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32**
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP45]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP45]], align 8
// CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP31]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 8
// CHECK10-NEXT: store i64 8, i64* [[TMP46]], align 8 // CHECK10-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 // CHECK10-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK10-NEXT: store i8* null, i8** [[TMP47]], align 8 // CHECK10-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2 // CHECK10-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK10-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32** // CHECK10-NEXT: [[TMP49:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP49]], align 8 // CHECK10-NEXT: store i32 [[TMP49]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 2 // CHECK10-NEXT: [[TMP50:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK10-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32** // CHECK10-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP50]], 0
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP51]], align 8 // CHECK10-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK10-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 2 // CHECK10-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK10-NEXT: store i64 [[TMP35]], i64* [[TMP52]], align 8 // CHECK10-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2 // CHECK10-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4
// CHECK10-NEXT: store i8* null, i8** [[TMP53]], align 8 // CHECK10-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP51]], 1
// CHECK10-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 // CHECK10-NEXT: [[TMP52:%.*]] = zext i32 [[ADD14]] to i64
// CHECK10-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP52]])
// CHECK10-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP53:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP47]], i8** [[TMP48]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP57:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK10-NEXT: store i32 [[TMP57]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10-NEXT: br i1 [[TMP54]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]
// CHECK10-NEXT: [[TMP58:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK10: omp_offload.failed15:
// CHECK10-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP58]], 0 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i64 [[TMP30]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT16]]
// CHECK10-NEXT: [[SUB14:%.*]] = sub nsw i32 [[DIV13]], 1 // CHECK10: omp_offload.cont16:
// CHECK10-NEXT: store i32 [[SUB14]], i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: [[TMP55:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_11]], align 4 // CHECK10-NEXT: [[CONV18:%.*]] = bitcast i64* [[N_CASTED17]] to i32*
// CHECK10-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP59]], 1 // CHECK10-NEXT: store i32 [[TMP55]], i32* [[CONV18]], align 4
// CHECK10-NEXT: [[TMP60:%.*]] = zext i32 [[ADD15]] to i64 // CHECK10-NEXT: [[TMP56:%.*]] = load i64, i64* [[N_CASTED17]], align 8
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP60]]) // CHECK10-NEXT: [[TMP57:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK10-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP54]], i8** [[TMP55]], i64* [[TMP56]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP58:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK10-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0 // CHECK10-NEXT: [[TMP59:%.*]] = bitcast i8** [[TMP58]] to i64*
// CHECK10-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] // CHECK10-NEXT: store i64 [[TMP56]], i64* [[TMP59]], align 8
// CHECK10: omp_offload.failed16: // CHECK10-NEXT: [[TMP60:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i64 [[TMP34]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]] // CHECK10-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i64*
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT17]] // CHECK10-NEXT: store i64 [[TMP56]], i64* [[TMP61]], align 8
// CHECK10: omp_offload.cont17: // CHECK10-NEXT: [[TMP62:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 0
// CHECK10-NEXT: [[TMP63:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: store i8* null, i8** [[TMP62]], align 8
// CHECK10-NEXT: [[CONV19:%.*]] = bitcast i64* [[N_CASTED18]] to i32* // CHECK10-NEXT: [[TMP63:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1
// CHECK10-NEXT: store i32 [[TMP63]], i32* [[CONV19]], align 4 // CHECK10-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i64*
// CHECK10-NEXT: [[TMP64:%.*]] = load i64, i64* [[N_CASTED18]], align 8 // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP64]], align 8
// CHECK10-NEXT: [[TMP65:%.*]] = mul nuw i64 [[TMP1]], 4 // CHECK10-NEXT: [[TMP65:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 1
// CHECK10-NEXT: [[TMP66:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK10-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i64*
// CHECK10-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i64* // CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP66]], align 8
// CHECK10-NEXT: store i64 [[TMP64]], i64* [[TMP67]], align 8 // CHECK10-NEXT: [[TMP67:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 1
// CHECK10-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK10-NEXT: store i8* null, i8** [[TMP67]], align 8
// CHECK10-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i64* // CHECK10-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2
// CHECK10-NEXT: store i64 [[TMP64]], i64* [[TMP69]], align 8 // CHECK10-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32**
// CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 0 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP69]], align 8
// CHECK10-NEXT: store i64 4, i64* [[TMP70]], align 8 // CHECK10-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 2
// CHECK10-NEXT: [[TMP71:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK10-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32**
// CHECK10-NEXT: store i8* null, i8** [[TMP71]], align 8 // CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP71]], align 8
// CHECK10-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 // CHECK10-NEXT: store i64 [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 2), align 8
// CHECK10-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i64* // CHECK10-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 2
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP73]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP72]], align 8
// CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 // CHECK10-NEXT: [[TMP73:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK10-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i64* // CHECK10-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK10-NEXT: store i64 [[TMP1]], i64* [[TMP75]], align 8 // CHECK10-NEXT: [[TMP75:%.*]] = load i32, i32* [[N]], align 4
// CHECK10-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 1 // CHECK10-NEXT: store i32 [[TMP75]], i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK10-NEXT: store i64 8, i64* [[TMP76]], align 8 // CHECK10-NEXT: [[TMP76:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK10-NEXT: [[TMP77:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 // CHECK10-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP76]], 0
// CHECK10-NEXT: store i8* null, i8** [[TMP77]], align 8 // CHECK10-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK10-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2 // CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK10-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32** // CHECK10-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP79]], align 8 // CHECK10-NEXT: [[TMP77:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK10-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 // CHECK10-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP77]], 1
// CHECK10-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32** // CHECK10-NEXT: [[TMP78:%.*]] = zext i32 [[ADD28]] to i64
// CHECK10-NEXT: store i32* [[VLA]], i32** [[TMP81]], align 8 // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP78]])
// CHECK10-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 2 // CHECK10-NEXT: [[TMP79:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP73]], i8** [[TMP74]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: store i64 [[TMP65]], i64* [[TMP82]], align 8 // CHECK10-NEXT: [[TMP80:%.*]] = icmp ne i32 [[TMP79]], 0
// CHECK10-NEXT: [[TMP83:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 // CHECK10-NEXT: br i1 [[TMP80]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK10-NEXT: store i8* null, i8** [[TMP83]], align 8 // CHECK10: omp_offload.failed29:
// CHECK10-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i64 [[TMP56]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: [[TMP85:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK10-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES23]], i32 0, i32 0 // CHECK10: omp_offload.cont30:
// CHECK10-NEXT: [[TMP87:%.*]] = load i32, i32* [[N]], align 4 // CHECK10-NEXT: [[TMP81:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: store i32 [[TMP87]], i32* [[DOTCAPTURE_EXPR_25]], align 4 // CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP81]])
// CHECK10-NEXT: [[TMP88:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_25]], align 4
// CHECK10-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP88]], 0
// CHECK10-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK10-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1
// CHECK10-NEXT: store i32 [[SUB29]], i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[TMP89:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_26]], align 4
// CHECK10-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP89]], 1
// CHECK10-NEXT: [[TMP90:%.*]] = zext i32 [[ADD30]] to i64
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP90]])
// CHECK10-NEXT: [[TMP91:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP84]], i8** [[TMP85]], i64* [[TMP86]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP92:%.*]] = icmp ne i32 [[TMP91]], 0
// CHECK10-NEXT: br i1 [[TMP92]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK10: omp_offload.failed31:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i64 [[TMP64]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT32]]
// CHECK10: omp_offload.cont32:
// CHECK10-NEXT: [[TMP93:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK10-NEXT: [[CALL:%.*]] = call signext i32 @_Z5tmainIiLi10EEiT_(i32 signext [[TMP93]])
// CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK10-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: [[TMP94:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK10-NEXT: [[TMP82:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP94]]) // CHECK10-NEXT: call void @llvm.stackrestore(i8* [[TMP82]])
// CHECK10-NEXT: [[TMP95:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK10-NEXT: [[TMP83:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP95]] // CHECK10-NEXT: ret i32 [[TMP83]]
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK10-SAME: (i64 [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
▲ Show 20 Lines • Show All 252 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 // CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK10-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32* // CHECK10-NEXT: [[CONV1:%.*]] = bitcast i64* [[DOTCAPTURE_EXPR__CASTED]] to i32*
// CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4 // CHECK10-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK10-NEXT: [[TMP4:%.*]] = load i64, i64* [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*, i64)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]], i64 [[TMP4]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i64 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 // CHECK10-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 140 Lines • ▼ Show 20 Lines
// CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK10-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK10-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP4]], align 8
// CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK10-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK10-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK10-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed: // CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont: // CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK10-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 8
// CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK10-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK10-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK10-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK10-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK10: omp_offload.failed5: // CHECK10: omp_offload.failed5:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK10: omp_offload.cont6: // CHECK10: omp_offload.cont6:
// CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK10-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 8
// CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK10-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 8
// CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK10-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP22]], align 8 // CHECK10-NEXT: store i8* null, i8** [[TMP22]], align 8
// CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK10-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK10-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK10-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK10-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK10-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK10-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK10-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK10-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK10: omp_offload.failed11: // CHECK10: omp_offload.failed11:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK10: omp_offload.cont12: // CHECK10: omp_offload.cont12:
// CHECK10-NEXT: ret i32 0 // CHECK10-NEXT: ret i32 0
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Lines
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 // CHECK10-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK10-NEXT: ret void // CHECK10-NEXT: ret void
// //
// //
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT: entry: // CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 // CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 // CHECK10-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 94 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK11-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_CASTED16:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS17:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS18:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS19:%.*]] = alloca [3 x i8*], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [3 x i8*], align 4 // CHECK11-NEXT: [[_TMP20:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOFFLOAD_SIZES21:%.*]] = alloca [3 x i64], align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_21:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP22:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK11-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK11-NEXT: store i32 100, i32* [[N]], align 4 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK11-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK11-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK11-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK11-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK11-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK11-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK11-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK11-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK11-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK11-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK11-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK11-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK11-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK11-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK11-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK11-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK11-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK11-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK11-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK11-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK11-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK11-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK11-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK11-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK11-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK11-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK11-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK11-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK11-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK11-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK11-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK11-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK11-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP53]])
// CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK11-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK11: omp_offload.failed14:
// CHECK11-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK11-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK11: omp_offload.cont15:
// CHECK11-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK11-NEXT: store i32 [[TMP56]], i32* [[N_CASTED16]], align 4
// CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK11-NEXT: [[TMP57:%.*]] = load i32, i32* [[N_CASTED16]], align 4
// CHECK11-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK11-NEXT: [[TMP58:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP61]]) // CHECK11-NEXT: [[TMP59:%.*]] = sext i32 [[TMP58]] to i64
// CHECK11-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP60:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK11-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK11-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i32*
// CHECK11-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK11-NEXT: store i32 [[TMP57]], i32* [[TMP61]], align 4
// CHECK11: omp_offload.failed15: // CHECK11-NEXT: [[TMP62:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK11-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32*
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK11-NEXT: store i32 [[TMP57]], i32* [[TMP63]], align 4
// CHECK11: omp_offload.cont16: // CHECK11-NEXT: [[TMP64:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 0
// CHECK11-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP64]], align 4
// CHECK11-NEXT: store i32 [[TMP64]], i32* [[N_CASTED17]], align 4 // CHECK11-NEXT: [[TMP65:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 1
// CHECK11-NEXT: [[TMP65:%.*]] = load i32, i32* [[N_CASTED17]], align 4 // CHECK11-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK11-NEXT: [[TMP66:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP66]], align 4
// CHECK11-NEXT: [[TMP67:%.*]] = sext i32 [[TMP66]] to i64 // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 1
// CHECK11-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK11-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i32*
// CHECK11-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32* // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP68]], align 4
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[TMP69]], align 4 // CHECK11-NEXT: [[TMP69:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 1
// CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK11-NEXT: store i8* null, i8** [[TMP69]], align 4
// CHECK11-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32* // CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP65]], i32* [[TMP71]], align 4 // CHECK11-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32**
// CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP71]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP72]], align 4 // CHECK11-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 2
// CHECK11-NEXT: [[TMP73:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0 // CHECK11-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32**
// CHECK11-NEXT: store i8* null, i8** [[TMP73]], align 4 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP73]], align 4
// CHECK11-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1 // CHECK11-NEXT: store i64 [[TMP59]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 2), align 4
// CHECK11-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32* // CHECK11-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP75]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP74]], align 4
// CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1 // CHECK11-NEXT: [[TMP75:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK11-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32* // CHECK11-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP77]], align 4 // CHECK11-NEXT: [[TMP77:%.*]] = load i32, i32* [[N]], align 4
// CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 1 // CHECK11-NEXT: store i32 [[TMP77]], i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK11-NEXT: store i64 4, i64* [[TMP78]], align 4 // CHECK11-NEXT: [[TMP78:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1 // CHECK11-NEXT: [[SUB23:%.*]] = sub nsw i32 [[TMP78]], 0
// CHECK11-NEXT: store i8* null, i8** [[TMP79]], align 4 // CHECK11-NEXT: [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
// CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2 // CHECK11-NEXT: [[SUB25:%.*]] = sub nsw i32 [[DIV24]], 1
// CHECK11-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32** // CHECK11-NEXT: store i32 [[SUB25]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP81]], align 4 // CHECK11-NEXT: [[TMP79:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2 // CHECK11-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP79]], 1
// CHECK11-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i32** // CHECK11-NEXT: [[TMP80:%.*]] = zext i32 [[ADD26]] to i64
// CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP83]], align 4 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP80]])
// CHECK11-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 2 // CHECK11-NEXT: [[TMP81:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP75]], i8** [[TMP76]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: store i64 [[TMP67]], i64* [[TMP84]], align 4 // CHECK11-NEXT: [[TMP82:%.*]] = icmp ne i32 [[TMP81]], 0
// CHECK11-NEXT: [[TMP85:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2 // CHECK11-NEXT: br i1 [[TMP82]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK11-NEXT: store i8* null, i8** [[TMP85]], align 4 // CHECK11: omp_offload.failed27:
// CHECK11-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i32 [[TMP57]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: [[TMP87:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK11-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK11: omp_offload.cont28:
// CHECK11-NEXT: [[TMP89:%.*]] = load i32, i32* [[N]], align 4 // CHECK11-NEXT: [[TMP83:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP89]], i32* [[DOTCAPTURE_EXPR_23]], align 4 // CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP83]])
// CHECK11-NEXT: [[TMP90:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK11-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP90]], 0
// CHECK11-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK11-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK11-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[TMP91:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP91]], 1
// CHECK11-NEXT: [[TMP92:%.*]] = zext i32 [[ADD28]] to i64
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP92]])
// CHECK11-NEXT: [[TMP93:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP86]], i8** [[TMP87]], i64* [[TMP88]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP94:%.*]] = icmp ne i32 [[TMP93]], 0
// CHECK11-NEXT: br i1 [[TMP94]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK11: omp_offload.failed29:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i32 [[TMP65]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK11: omp_offload.cont30:
// CHECK11-NEXT: [[TMP95:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP95]])
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP96:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP84:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP96]]) // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP84]])
// CHECK11-NEXT: [[TMP97:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP85:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP97]] // CHECK11-NEXT: ret i32 [[TMP85]]
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 246 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 138 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK11-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK11-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed: // CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont: // CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK11-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK11: omp_offload.failed5: // CHECK11: omp_offload.failed5:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK11: omp_offload.cont6: // CHECK11: omp_offload.cont6:
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK11-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK11-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0
// CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK11-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK11-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK11-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK11-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK11-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK11-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK11: omp_offload.failed11: // CHECK11: omp_offload.failed11:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK11: omp_offload.cont12: // CHECK11: omp_offload.cont12:
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK11-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK11-NEXT: ret void // CHECK11-NEXT: ret void
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 93 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 // CHECK12-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4
// CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK12-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 // CHECK12-NEXT: [[_TMP7:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_8:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_CASTED16:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[N_CASTED17:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS17:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_BASEPTRS18:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_PTRS18:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_PTRS19:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS19:%.*]] = alloca [3 x i8*], align 4
// CHECK12-NEXT: [[DOTOFFLOAD_MAPPERS20:%.*]] = alloca [3 x i8*], align 4 // CHECK12-NEXT: [[_TMP20:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOFFLOAD_SIZES21:%.*]] = alloca [3 x i64], align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_21:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[_TMP22:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR_22:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_23:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 // CHECK12-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 // CHECK12-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4
// CHECK12-NEXT: store i32 100, i32* [[N]], align 4 // CHECK12-NEXT: store i32 100, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() // CHECK12-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave()
// CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 // CHECK12-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 // CHECK12-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* // CHECK12-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP7]], align 4
// CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32* // CHECK12-NEXT: [[TMP9:%.*]] = bitcast i8** [[TMP8]] to i32*
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[TMP9]], align 4
// CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK12-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP10]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP11]], align 4 // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32*
// CHECK12-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 // CHECK12-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to i32*
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP14]], align 4
// CHECK12-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP15]], align 4
// CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK12-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: store i64 4, i64* [[TMP16]], align 4 // CHECK12-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i32**
// CHECK12-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP17]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP17]], align 4 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to i32**
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK12-NEXT: store i64 [[TMP5]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i32** // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP21]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP20]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK12-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: store i64 [[TMP5]], i64* [[TMP22]], align 4 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK12-NEXT: [[TMP23:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP23]], align 4 // CHECK12-NEXT: store i32 [[TMP23]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK12-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK12-NEXT: [[TMP27:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP27]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP28]], 0
// CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK12-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 // CHECK12-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 // CHECK12-NEXT: [[TMP25:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4
// CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP29]], 1 // CHECK12-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK12-NEXT: [[TMP30:%.*]] = zext i32 [[ADD]] to i64 // CHECK12-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP30]]) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP26]])
// CHECK12-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP24]], i8** [[TMP25]], i64* [[TMP26]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP27:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, i32 3, i8** [[TMP21]], i8** [[TMP22]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK12-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0
// CHECK12-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP28]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP33:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP29:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32 [[TMP33]], i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: store i32 [[TMP29]], i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP34:%.*]] = load i32, i32* [[N_CASTED3]], align 4 // CHECK12-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_CASTED3]], align 4
// CHECK12-NEXT: [[TMP35:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: [[TMP31:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK12-NEXT: [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
// CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK12-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP38]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP34]], align 4
// CHECK12-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK12-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK12-NEXT: store i32 [[TMP34]], i32* [[TMP40]], align 4 // CHECK12-NEXT: store i32 [[TMP30]], i32* [[TMP36]], align 4
// CHECK12-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK12-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 // CHECK12-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK12-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP39]], align 4
// CHECK12-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK12-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP41]], align 4
// CHECK12-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 // CHECK12-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK12-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP44]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP44]], align 4
// CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 // CHECK12-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK12-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32**
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP46]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP46]], align 4
// CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP32]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 2), align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK12-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 // CHECK12-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK12-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 // CHECK12-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK12-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to i32** // CHECK12-NEXT: [[TMP50:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP50]], align 4 // CHECK12-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 // CHECK12-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_8]], align 4
// CHECK12-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32** // CHECK12-NEXT: [[SUB10:%.*]] = sub nsw i32 [[TMP51]], 0
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP52]], align 4 // CHECK12-NEXT: [[DIV11:%.*]] = sdiv i32 [[SUB10]], 1
// CHECK12-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 // CHECK12-NEXT: [[SUB12:%.*]] = sub nsw i32 [[DIV11]], 1
// CHECK12-NEXT: store i64 [[TMP36]], i64* [[TMP53]], align 4 // CHECK12-NEXT: store i32 [[SUB12]], i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 // CHECK12-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4
// CHECK12-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK12-NEXT: [[ADD13:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK12-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 // CHECK12-NEXT: [[TMP53:%.*]] = zext i32 [[ADD13]] to i64
// CHECK12-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP53]])
// CHECK12-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP58:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0
// CHECK12-NEXT: store i32 [[TMP58]], i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED14:%.*]], label [[OMP_OFFLOAD_CONT15:%.*]]
// CHECK12-NEXT: [[TMP59:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 // CHECK12: omp_offload.failed14:
// CHECK12-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP59]], 0 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i32 [[TMP30]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT15]]
// CHECK12-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 // CHECK12: omp_offload.cont15:
// CHECK12-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: [[TMP56:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP60:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 // CHECK12-NEXT: store i32 [[TMP56]], i32* [[N_CASTED16]], align 4
// CHECK12-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP60]], 1 // CHECK12-NEXT: [[TMP57:%.*]] = load i32, i32* [[N_CASTED16]], align 4
// CHECK12-NEXT: [[TMP61:%.*]] = zext i32 [[ADD14]] to i64 // CHECK12-NEXT: [[TMP58:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP61]]) // CHECK12-NEXT: [[TMP59:%.*]] = sext i32 [[TMP58]] to i64
// CHECK12-NEXT: [[TMP62:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, i32 3, i8** [[TMP55]], i8** [[TMP56]], i64* [[TMP57]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP60:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK12-NEXT: [[TMP63:%.*]] = icmp ne i32 [[TMP62]], 0 // CHECK12-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i32*
// CHECK12-NEXT: br i1 [[TMP63]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] // CHECK12-NEXT: store i32 [[TMP57]], i32* [[TMP61]], align 4
// CHECK12: omp_offload.failed15: // CHECK12-NEXT: [[TMP62:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i32 [[TMP34]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]] // CHECK12-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32*
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT16]] // CHECK12-NEXT: store i32 [[TMP57]], i32* [[TMP63]], align 4
// CHECK12: omp_offload.cont16: // CHECK12-NEXT: [[TMP64:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 0
// CHECK12-NEXT: [[TMP64:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP64]], align 4
// CHECK12-NEXT: store i32 [[TMP64]], i32* [[N_CASTED17]], align 4 // CHECK12-NEXT: [[TMP65:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 1
// CHECK12-NEXT: [[TMP65:%.*]] = load i32, i32* [[N_CASTED17]], align 4 // CHECK12-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK12-NEXT: [[TMP66:%.*]] = mul nuw i32 [[TMP0]], 4 // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP66]], align 4
// CHECK12-NEXT: [[TMP67:%.*]] = sext i32 [[TMP66]] to i64 // CHECK12-NEXT: [[TMP67:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 1
// CHECK12-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK12-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i32*
// CHECK12-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32* // CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP68]], align 4
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[TMP69]], align 4 // CHECK12-NEXT: [[TMP69:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 1
// CHECK12-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK12-NEXT: store i8* null, i8** [[TMP69]], align 4
// CHECK12-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32* // CHECK12-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP65]], i32* [[TMP71]], align 4 // CHECK12-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32**
// CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP71]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP72]], align 4 // CHECK12-NEXT: [[TMP72:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 2
// CHECK12-NEXT: [[TMP73:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 0 // CHECK12-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32**
// CHECK12-NEXT: store i8* null, i8** [[TMP73]], align 4 // CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP73]], align 4
// CHECK12-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 1 // CHECK12-NEXT: store i64 [[TMP59]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 2), align 4
// CHECK12-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32* // CHECK12-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS19]], i32 0, i32 2
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP75]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP74]], align 4
// CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 1 // CHECK12-NEXT: [[TMP75:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS17]], i32 0, i32 0
// CHECK12-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32* // CHECK12-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS18]], i32 0, i32 0
// CHECK12-NEXT: store i32 [[TMP0]], i32* [[TMP77]], align 4 // CHECK12-NEXT: [[TMP77:%.*]] = load i32, i32* [[N]], align 4
// CHECK12-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 1 // CHECK12-NEXT: store i32 [[TMP77]], i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK12-NEXT: store i64 4, i64* [[TMP78]], align 4 // CHECK12-NEXT: [[TMP78:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_21]], align 4
// CHECK12-NEXT: [[TMP79:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 1 // CHECK12-NEXT: [[SUB23:%.*]] = sub nsw i32 [[TMP78]], 0
// CHECK12-NEXT: store i8* null, i8** [[TMP79]], align 4 // CHECK12-NEXT: [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
// CHECK12-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 2 // CHECK12-NEXT: [[SUB25:%.*]] = sub nsw i32 [[DIV24]], 1
// CHECK12-NEXT: [[TMP81:%.*]] = bitcast i8** [[TMP80]] to i32** // CHECK12-NEXT: store i32 [[SUB25]], i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP81]], align 4 // CHECK12-NEXT: [[TMP79:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_22]], align 4
// CHECK12-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 2 // CHECK12-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP79]], 1
// CHECK12-NEXT: [[TMP83:%.*]] = bitcast i8** [[TMP82]] to i32** // CHECK12-NEXT: [[TMP80:%.*]] = zext i32 [[ADD26]] to i64
// CHECK12-NEXT: store i32* [[VLA]], i32** [[TMP83]], align 4 // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP80]])
// CHECK12-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 2 // CHECK12-NEXT: [[TMP81:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP75]], i8** [[TMP76]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.5, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: store i64 [[TMP67]], i64* [[TMP84]], align 4 // CHECK12-NEXT: [[TMP82:%.*]] = icmp ne i32 [[TMP81]], 0
// CHECK12-NEXT: [[TMP85:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS20]], i32 0, i32 2 // CHECK12-NEXT: br i1 [[TMP82]], label [[OMP_OFFLOAD_FAILED27:%.*]], label [[OMP_OFFLOAD_CONT28:%.*]]
// CHECK12-NEXT: store i8* null, i8** [[TMP85]], align 4 // CHECK12: omp_offload.failed27:
// CHECK12-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS18]], i32 0, i32 0 // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i32 [[TMP57]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: [[TMP87:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS19]], i32 0, i32 0 // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT28]]
// CHECK12-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES21]], i32 0, i32 0 // CHECK12: omp_offload.cont28:
// CHECK12-NEXT: [[TMP89:%.*]] = load i32, i32* [[N]], align 4 // CHECK12-NEXT: [[TMP83:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP89]], i32* [[DOTCAPTURE_EXPR_23]], align 4 // CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP83]])
// CHECK12-NEXT: [[TMP90:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_23]], align 4
// CHECK12-NEXT: [[SUB25:%.*]] = sub nsw i32 [[TMP90]], 0
// CHECK12-NEXT: [[DIV26:%.*]] = sdiv i32 [[SUB25]], 1
// CHECK12-NEXT: [[SUB27:%.*]] = sub nsw i32 [[DIV26]], 1
// CHECK12-NEXT: store i32 [[SUB27]], i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[TMP91:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_24]], align 4
// CHECK12-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP91]], 1
// CHECK12-NEXT: [[TMP92:%.*]] = zext i32 [[ADD28]] to i64
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP92]])
// CHECK12-NEXT: [[TMP93:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, i32 3, i8** [[TMP86]], i8** [[TMP87]], i64* [[TMP88]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP94:%.*]] = icmp ne i32 [[TMP93]], 0
// CHECK12-NEXT: br i1 [[TMP94]], label [[OMP_OFFLOAD_FAILED29:%.*]], label [[OMP_OFFLOAD_CONT30:%.*]]
// CHECK12: omp_offload.failed29:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i32 [[TMP65]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT30]]
// CHECK12: omp_offload.cont30:
// CHECK12-NEXT: [[TMP95:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK12-NEXT: [[CALL:%.*]] = call i32 @_Z5tmainIiLi10EEiT_(i32 [[TMP95]])
// CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 // CHECK12-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK12-NEXT: [[TMP96:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK12-NEXT: [[TMP84:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP96]]) // CHECK12-NEXT: call void @llvm.stackrestore(i8* [[TMP84]])
// CHECK12-NEXT: [[TMP97:%.*]] = load i32, i32* [[RETVAL]], align 4 // CHECK12-NEXT: [[TMP85:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK12-NEXT: ret i32 [[TMP97]] // CHECK12-NEXT: ret i32 [[TMP85]]
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100
// CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK12-SAME: (i32 [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
▲ Show 20 Lines • Show All 246 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 // CHECK12-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4
// CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK12-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 // CHECK12-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4
// CHECK12-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK12-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*, i32)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]], i32 [[TMP4]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..3 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i32 [[VLA:%.*]], i32* nonnull align 4 dereferenceable(4) [[A:%.*]], i32 [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 138 Lines • ▼ Show 20 Lines
// CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]** // CHECK12-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP3]], align 4
// CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP4]], align 4
// CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK12-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.6, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 // CHECK12-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0
// CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK12-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK12: omp_offload.failed: // CHECK12: omp_offload.failed:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK12: omp_offload.cont: // CHECK12: omp_offload.cont:
// CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]** // CHECK12-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP10]], align 4
// CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]** // CHECK12-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP12]], align 4
// CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK12-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP13]], align 4
// CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK12-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK12-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.8, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.9, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, i32 1, i8** [[TMP14]], i8** [[TMP15]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK12-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0
// CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK12-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK12: omp_offload.failed5: // CHECK12: omp_offload.failed5:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK12: omp_offload.cont6: // CHECK12: omp_offload.cont6:
// CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]** // CHECK12-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP19]], align 4
// CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]** // CHECK12-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to [10 x i32]**
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP21]], align 4
// CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0 // CHECK12-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i32 0, i32 0
// CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4 // CHECK12-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK12-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK12-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) // CHECK12-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10)
// CHECK12-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.11, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.12, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) // CHECK12-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, i32 1, i8** [[TMP23]], i8** [[TMP24]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.14, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.15, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1)
// CHECK12-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK12-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK12-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK12-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK12: omp_offload.failed11: // CHECK12: omp_offload.failed11:
// CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]] // CHECK12-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89([10 x i32]* [[A]]) #[[ATTR3]]
// CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK12-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK12: omp_offload.cont12: // CHECK12: omp_offload.cont12:
// CHECK12-NEXT: ret i32 0 // CHECK12-NEXT: ret i32 0
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..5 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..7 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 // CHECK12-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: ([10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 // CHECK12-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4
// CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]]) // CHECK12-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP0]])
// CHECK12-NEXT: ret void // CHECK12-NEXT: ret void
// //
// //
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..10 // CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { // CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [10 x i32]* nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT: entry: // CHECK12-NEXT: entry:
// CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 // CHECK12-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 // CHECK12-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4
// CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK12-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 1,325 Lines • Show Last 20 Lines

clang/test/OpenMP/teams_firstprivate_codegen.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 3,319 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[VLA1_ADDR:%.*]] = alloca ppc_fp128*, align 8 // CHECK17-NEXT: [[VLA1_ADDR:%.*]] = alloca ppc_fp128*, align 8
// CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [8 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [8 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [8 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [8 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [8 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [8 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [8 x i64], align 8
// CHECK17-NEXT: store float* [[A]], float** [[A_ADDR]], align 8 // CHECK17-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK17-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8 // CHECK17-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8
// CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK17-NEXT: store ppc_fp128* [[VLA1]], ppc_fp128** [[VLA1_ADDR]], align 8 // CHECK17-NEXT: store ppc_fp128* [[VLA1]], ppc_fp128** [[VLA1_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK17-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK17-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64 // CHECK17-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
Show All 15 Lines
// CHECK17-NEXT: [[TMP13:%.*]] = mul nuw i64 [[TMP3]], [[TMP5]] // CHECK17-NEXT: [[TMP13:%.*]] = mul nuw i64 [[TMP3]], [[TMP5]]
// CHECK17-NEXT: [[TMP14:%.*]] = mul nuw i64 [[TMP13]], 8 // CHECK17-NEXT: [[TMP14:%.*]] = mul nuw i64 [[TMP13]], 8
// CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK17-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float**
// CHECK17-NEXT: store float* [[TMP8]], float** [[TMP16]], align 8 // CHECK17-NEXT: store float* [[TMP8]], float** [[TMP16]], align 8
// CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to float** // CHECK17-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to float**
// CHECK17-NEXT: store float* [[TMP8]], float** [[TMP18]], align 8 // CHECK17-NEXT: store float* [[TMP8]], float** [[TMP18]], align 8
// CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK17-NEXT: store i64 0, i64* [[TMP19]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK17-NEXT: store i8* null, i8** [[TMP20]], align 8 // CHECK17-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.St**
// CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK17-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP21]], align 8
// CHECK17-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.St** // CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK17-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP22]], align 8 // CHECK17-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.St**
// CHECK17-NEXT: [[TMP23:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK17-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP23]], align 8
// CHECK17-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to %struct.St** // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK17-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP24]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK17-NEXT: [[TMP25:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK17-NEXT: [[TMP25:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: store i64 0, i64* [[TMP25]], align 8 // CHECK17-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK17-NEXT: store i64 [[TMP1]], i64* [[TMP26]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64* // CHECK17-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK17-NEXT: store i64 [[TMP1]], i64* [[TMP28]], align 8 // CHECK17-NEXT: store i64 [[TMP1]], i64* [[TMP28]], align 8
// CHECK17-NEXT: [[TMP29:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK17-NEXT: [[TMP29:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK17-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64* // CHECK17-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK17-NEXT: store i64 [[TMP1]], i64* [[TMP30]], align 8 // CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK17-NEXT: [[TMP31:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK17-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to ppc_fp128**
// CHECK17-NEXT: store i64 8, i64* [[TMP31]], align 8 // CHECK17-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP31]], align 8
// CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK17-NEXT: store i8* null, i8** [[TMP32]], align 8 // CHECK17-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to ppc_fp128**
// CHECK17-NEXT: [[TMP33:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK17-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP33]], align 8
// CHECK17-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to ppc_fp128** // CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK17-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP34]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK17-NEXT: [[TMP35:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK17-NEXT: [[TMP35:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK17-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to ppc_fp128** // CHECK17-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i64*
// CHECK17-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP36]], align 8 // CHECK17-NEXT: store i64 [[TMP3]], i64* [[TMP36]], align 8
// CHECK17-NEXT: [[TMP37:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK17-NEXT: [[TMP37:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK17-NEXT: store i64 0, i64* [[TMP37]], align 8 // CHECK17-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK17-NEXT: [[TMP38:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK17-NEXT: store i64 [[TMP3]], i64* [[TMP38]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP38]], align 8 // CHECK17-NEXT: [[TMP39:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK17-NEXT: [[TMP39:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK17-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK17-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64* // CHECK17-NEXT: [[TMP40:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK17-NEXT: store i64 [[TMP3]], i64* [[TMP40]], align 8 // CHECK17-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i64*
// CHECK17-NEXT: [[TMP41:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP41]], align 8
// CHECK17-NEXT: [[TMP42:%.*]] = bitcast i8** [[TMP41]] to i64* // CHECK17-NEXT: [[TMP42:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK17-NEXT: store i64 [[TMP3]], i64* [[TMP42]], align 8 // CHECK17-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64*
// CHECK17-NEXT: [[TMP43:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP43]], align 8
// CHECK17-NEXT: store i64 8, i64* [[TMP43]], align 8 // CHECK17-NEXT: [[TMP44:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK17-NEXT: [[TMP44:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK17-NEXT: store i8* null, i8** [[TMP44]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP44]], align 8
// CHECK17-NEXT: [[TMP45:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK17-NEXT: [[TMP45:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK17-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i64* // CHECK17-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to double**
// CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP46]], align 8 // CHECK17-NEXT: store double* [[VLA]], double** [[TMP46]], align 8
// CHECK17-NEXT: [[TMP47:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK17-NEXT: [[TMP47:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK17-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i64* // CHECK17-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to double**
// CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP48]], align 8 // CHECK17-NEXT: store double* [[VLA]], double** [[TMP48]], align 8
// CHECK17-NEXT: [[TMP49:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK17-NEXT: store i64 [[TMP14]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_sizes, i32 0, i32 6), align 8
// CHECK17-NEXT: store i64 8, i64* [[TMP49]], align 8 // CHECK17-NEXT: [[TMP49:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6
// CHECK17-NEXT: [[TMP50:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5 // CHECK17-NEXT: store i8* null, i8** [[TMP49]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP50]], align 8 // CHECK17-NEXT: [[TMP50:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7
// CHECK17-NEXT: [[TMP51:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK17-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i64*
// CHECK17-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to double** // CHECK17-NEXT: store i64 [[TMP12]], i64* [[TMP51]], align 8
// CHECK17-NEXT: store double* [[VLA]], double** [[TMP52]], align 8 // CHECK17-NEXT: [[TMP52:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK17-NEXT: [[TMP53:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6 // CHECK17-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to i64*
// CHECK17-NEXT: [[TMP54:%.*]] = bitcast i8** [[TMP53]] to double** // CHECK17-NEXT: store i64 [[TMP12]], i64* [[TMP53]], align 8
// CHECK17-NEXT: store double* [[VLA]], double** [[TMP54]], align 8 // CHECK17-NEXT: [[TMP54:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7
// CHECK17-NEXT: [[TMP55:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK17-NEXT: store i8* null, i8** [[TMP54]], align 8
// CHECK17-NEXT: store i64 [[TMP14]], i64* [[TMP55]], align 8 // CHECK17-NEXT: [[TMP55:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP56:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6 // CHECK17-NEXT: [[TMP56:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: store i8* null, i8** [[TMP56]], align 8 // CHECK17-NEXT: [[TMP57:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152.region_id, i32 8, i8** [[TMP55]], i8** [[TMP56]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK17-NEXT: [[TMP57:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7 // CHECK17-NEXT: [[TMP58:%.*]] = icmp ne i32 [[TMP57]], 0
// CHECK17-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to i64* // CHECK17-NEXT: br i1 [[TMP58]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17-NEXT: store i64 [[TMP12]], i64* [[TMP58]], align 8
// CHECK17-NEXT: [[TMP59:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK17-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i64*
// CHECK17-NEXT: store i64 [[TMP12]], i64* [[TMP60]], align 8
// CHECK17-NEXT: [[TMP61:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7
// CHECK17-NEXT: store i64 4, i64* [[TMP61]], align 8
// CHECK17-NEXT: [[TMP62:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7
// CHECK17-NEXT: store i8* null, i8** [[TMP62]], align 8
// CHECK17-NEXT: [[TMP63:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP64:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP65:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK17-NEXT: [[TMP66:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152.region_id, i32 8, i8** [[TMP63]], i8** [[TMP64]], i64* [[TMP65]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK17-NEXT: [[TMP67:%.*]] = icmp ne i32 [[TMP66]], 0
// CHECK17-NEXT: br i1 [[TMP67]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152(float* [[TMP8]], %struct.St* [[TMP9]], i64 [[TMP1]], ppc_fp128* [[TMP10]], i64 [[TMP3]], i64 [[TMP5]], double* [[VLA]], i64 [[TMP12]]) #[[ATTR4:[0-9]+]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152(float* [[TMP8]], %struct.St* [[TMP9]], i64 [[TMP1]], ppc_fp128* [[TMP10]], i64 [[TMP3]], i64 [[TMP5]], double* [[VLA]], i64 [[TMP12]]) #[[ATTR4:[0-9]+]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: [[TMP68:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK17-NEXT: [[TMP59:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP68]]) // CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP59]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152
// CHECK17-SAME: (float* [[A:%.*]], %struct.St* [[S:%.*]], i64 [[VLA:%.*]], ppc_fp128* [[VLA1:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], double* nonnull align 8 dereferenceable(8) [[VLA26:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK17-SAME: (float* [[A:%.*]], %struct.St* [[S:%.*]], i64 [[VLA:%.*]], ppc_fp128* [[VLA1:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], double* nonnull align 8 dereferenceable(8) [[VLA26:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8 // CHECK17-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 8 // CHECK17-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 8
▲ Show 20 Lines • Show All 85 Lines • ▼ Show 20 Lines
// CHECK17-NEXT: [[VLA1_ADDR:%.*]] = alloca ppc_fp128*, align 8 // CHECK17-NEXT: [[VLA1_ADDR:%.*]] = alloca ppc_fp128*, align 8
// CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK17-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [10 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [10 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [10 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [10 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [10 x i8*], align 8 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [10 x i8*], align 8
// CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK17-NEXT: store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8 // CHECK17-NEXT: store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
// CHECK17-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8 // CHECK17-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8
// CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK17-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK17-NEXT: store ppc_fp128* [[VLA1]], ppc_fp128** [[VLA1_ADDR]], align 8 // CHECK17-NEXT: store ppc_fp128* [[VLA1]], ppc_fp128** [[VLA1_ADDR]], align 8
// CHECK17-NEXT: [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8 // CHECK17-NEXT: [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
// CHECK17-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK17-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK17-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 28 Lines
// CHECK17-NEXT: [[TMP20:%.*]] = sub i64 [[TMP18]], [[TMP19]] // CHECK17-NEXT: [[TMP20:%.*]] = sub i64 [[TMP18]], [[TMP19]]
// CHECK17-NEXT: [[TMP21:%.*]] = sdiv exact i64 [[TMP20]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK17-NEXT: [[TMP21:%.*]] = sdiv exact i64 [[TMP20]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.St** // CHECK17-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.St**
// CHECK17-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP23]], align 8 // CHECK17-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP23]], align 8
// CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to %struct.St** // CHECK17-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to %struct.St**
// CHECK17-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP25]], align 8 // CHECK17-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP25]], align 8
// CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK17-NEXT: store i64 0, i64* [[TMP26]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK17-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK17-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK17-NEXT: store i64 [[TMP1]], i64* [[TMP28]], align 8
// CHECK17-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK17-NEXT: [[TMP29:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK17-NEXT: store i64 [[TMP1]], i64* [[TMP29]], align 8 // CHECK17-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK17-NEXT: store i64 [[TMP1]], i64* [[TMP30]], align 8
// CHECK17-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK17-NEXT: [[TMP31:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK17-NEXT: store i64 [[TMP1]], i64* [[TMP31]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK17-NEXT: [[TMP32:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK17-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to ppc_fp128**
// CHECK17-NEXT: [[TMP33:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK17-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP33]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP34:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK17-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to ppc_fp128** // CHECK17-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to ppc_fp128**
// CHECK17-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP35]], align 8 // CHECK17-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP35]], align 8
// CHECK17-NEXT: [[TMP36:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK17-NEXT: [[TMP36:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK17-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to ppc_fp128** // CHECK17-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK17-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP37]], align 8 // CHECK17-NEXT: [[TMP37:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK17-NEXT: [[TMP38:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK17-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK17-NEXT: store i64 0, i64* [[TMP38]], align 8 // CHECK17-NEXT: store i64 [[TMP3]], i64* [[TMP38]], align 8
// CHECK17-NEXT: [[TMP39:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK17-NEXT: [[TMP39:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK17-NEXT: store i8* null, i8** [[TMP39]], align 8 // CHECK17-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK17-NEXT: [[TMP40:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK17-NEXT: store i64 [[TMP3]], i64* [[TMP40]], align 8
// CHECK17-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i64* // CHECK17-NEXT: [[TMP41:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK17-NEXT: store i64 [[TMP3]], i64* [[TMP41]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK17-NEXT: [[TMP42:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK17-NEXT: [[TMP42:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK17-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK17-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64*
// CHECK17-NEXT: store i64 [[TMP3]], i64* [[TMP43]], align 8 // CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP43]], align 8
// CHECK17-NEXT: [[TMP44:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK17-NEXT: [[TMP44:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK17-NEXT: store i64 8, i64* [[TMP44]], align 8 // CHECK17-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64*
// CHECK17-NEXT: [[TMP45:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP45]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP45]], align 8 // CHECK17-NEXT: [[TMP46:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK17-NEXT: [[TMP46:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK17-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK17-NEXT: [[TMP47:%.*]] = bitcast i8** [[TMP46]] to i64* // CHECK17-NEXT: [[TMP47:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP47]], align 8 // CHECK17-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to double**
// CHECK17-NEXT: [[TMP48:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK17-NEXT: store double* [[VLA]], double** [[TMP48]], align 8
// CHECK17-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i64* // CHECK17-NEXT: [[TMP49:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK17-NEXT: store i64 [[TMP5]], i64* [[TMP49]], align 8 // CHECK17-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to double**
// CHECK17-NEXT: [[TMP50:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK17-NEXT: store double* [[VLA]], double** [[TMP50]], align 8
// CHECK17-NEXT: store i64 8, i64* [[TMP50]], align 8 // CHECK17-NEXT: store i64 [[TMP14]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 5), align 8
// CHECK17-NEXT: [[TMP51:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK17-NEXT: [[TMP51:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK17-NEXT: store i8* null, i8** [[TMP51]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP51]], align 8
// CHECK17-NEXT: [[TMP52:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK17-NEXT: [[TMP52:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK17-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to double** // CHECK17-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to %struct.St**
// CHECK17-NEXT: store double* [[VLA]], double** [[TMP53]], align 8 // CHECK17-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP53]], align 8
// CHECK17-NEXT: [[TMP54:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK17-NEXT: [[TMP54:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK17-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to double** // CHECK17-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to i32**
// CHECK17-NEXT: store double* [[VLA]], double** [[TMP55]], align 8 // CHECK17-NEXT: store i32* [[A3]], i32** [[TMP55]], align 8
// CHECK17-NEXT: [[TMP56:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK17-NEXT: store i64 [[TMP21]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 6), align 8
// CHECK17-NEXT: store i64 [[TMP14]], i64* [[TMP56]], align 8 // CHECK17-NEXT: [[TMP56:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6
// CHECK17-NEXT: [[TMP57:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5 // CHECK17-NEXT: store i8* null, i8** [[TMP56]], align 8
// CHECK17-NEXT: store i8* null, i8** [[TMP57]], align 8 // CHECK17-NEXT: [[TMP57:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7
// CHECK17-NEXT: [[TMP58:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK17-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to %struct.St**
// CHECK17-NEXT: [[TMP59:%.*]] = bitcast i8** [[TMP58]] to %struct.St** // CHECK17-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP58]], align 8
// CHECK17-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP59]], align 8 // CHECK17-NEXT: [[TMP59:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK17-NEXT: [[TMP60:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6 // CHECK17-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i32**
// CHECK17-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i32** // CHECK17-NEXT: store i32* [[B2]], i32** [[TMP60]], align 8
// CHECK17-NEXT: store i32* [[A3]], i32** [[TMP61]], align 8 // CHECK17-NEXT: [[TMP61:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7
// CHECK17-NEXT: [[TMP62:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK17-NEXT: store i8* null, i8** [[TMP61]], align 8
// CHECK17-NEXT: store i64 [[TMP21]], i64* [[TMP62]], align 8 // CHECK17-NEXT: [[TMP62:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8
// CHECK17-NEXT: [[TMP63:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6 // CHECK17-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to %struct.St**
// CHECK17-NEXT: store i8* null, i8** [[TMP63]], align 8 // CHECK17-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP63]], align 8
// CHECK17-NEXT: [[TMP64:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7 // CHECK17-NEXT: [[TMP64:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 8
// CHECK17-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to %struct.St** // CHECK17-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32**
// CHECK17-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP65]], align 8 // CHECK17-NEXT: store i32* [[A3]], i32** [[TMP65]], align 8
// CHECK17-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7 // CHECK17-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 8
// CHECK17-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32** // CHECK17-NEXT: store i8* null, i8** [[TMP66]], align 8
// CHECK17-NEXT: store i32* [[B2]], i32** [[TMP67]], align 8 // CHECK17-NEXT: [[TMP67:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK17-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK17-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i64*
// CHECK17-NEXT: store i64 4, i64* [[TMP68]], align 8 // CHECK17-NEXT: store i64 [[TMP12]], i64* [[TMP68]], align 8
// CHECK17-NEXT: [[TMP69:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7 // CHECK17-NEXT: [[TMP69:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK17-NEXT: store i8* null, i8** [[TMP69]], align 8 // CHECK17-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i64*
// CHECK17-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8 // CHECK17-NEXT: store i64 [[TMP12]], i64* [[TMP70]], align 8
// CHECK17-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to %struct.St** // CHECK17-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 9
// CHECK17-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP71]], align 8 // CHECK17-NEXT: store i8* null, i8** [[TMP71]], align 8
// CHECK17-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 8 // CHECK17-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32** // CHECK17-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: store i32* [[A3]], i32** [[TMP73]], align 8 // CHECK17-NEXT: [[TMP74:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144.region_id, i32 10, i8** [[TMP72]], i8** [[TMP73]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK17-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK17-NEXT: [[TMP75:%.*]] = icmp ne i32 [[TMP74]], 0
// CHECK17-NEXT: store i64 4, i64* [[TMP74]], align 8 // CHECK17-NEXT: br i1 [[TMP75]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17-NEXT: [[TMP75:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 8
// CHECK17-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK17-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK17-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK17-NEXT: store i64 [[TMP12]], i64* [[TMP77]], align 8
// CHECK17-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK17-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK17-NEXT: store i64 [[TMP12]], i64* [[TMP79]], align 8
// CHECK17-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK17-NEXT: store i64 4, i64* [[TMP80]], align 8
// CHECK17-NEXT: [[TMP81:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 9
// CHECK17-NEXT: store i8* null, i8** [[TMP81]], align 8
// CHECK17-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK17-NEXT: [[TMP84:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK17-NEXT: [[TMP85:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144.region_id, i32 10, i8** [[TMP82]], i8** [[TMP83]], i64* [[TMP84]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK17-NEXT: [[TMP86:%.*]] = icmp ne i32 [[TMP85]], 0
// CHECK17-NEXT: br i1 [[TMP86]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK17: omp_offload.failed: // CHECK17: omp_offload.failed:
// CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144(%struct.St* [[TMP9]], i64 [[TMP1]], ppc_fp128* [[TMP10]], i64 [[TMP3]], i64 [[TMP5]], double* [[VLA]], %struct.St* [[THIS1]], i64 [[TMP12]]) #[[ATTR4]] // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144(%struct.St* [[TMP9]], i64 [[TMP1]], ppc_fp128* [[TMP10]], i64 [[TMP3]], i64 [[TMP5]], double* [[VLA]], %struct.St* [[THIS1]], i64 [[TMP12]]) #[[ATTR4]]
// CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK17: omp_offload.cont: // CHECK17: omp_offload.cont:
// CHECK17-NEXT: [[TMP87:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK17-NEXT: [[TMP76:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP87]]) // CHECK17-NEXT: call void @llvm.stackrestore(i8* [[TMP76]])
// CHECK17-NEXT: ret void // CHECK17-NEXT: ret void
// //
// //
// CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144
// CHECK17-SAME: (%struct.St* [[S:%.*]], i64 [[VLA:%.*]], ppc_fp128* [[VLA1:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], double* nonnull align 8 dereferenceable(8) [[VLA26:%.*]], %struct.St* [[THIS:%.*]], i64 [[N:%.*]]) #[[ATTR2]] { // CHECK17-SAME: (%struct.St* [[S:%.*]], i64 [[VLA:%.*]], ppc_fp128* [[VLA1:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], double* nonnull align 8 dereferenceable(8) [[VLA26:%.*]], %struct.St* [[THIS:%.*]], i64 [[N:%.*]]) #[[ATTR2]] {
// CHECK17-NEXT: entry: // CHECK17-NEXT: entry:
// CHECK17-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 8 // CHECK17-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK17-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 106 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[VLA1_ADDR:%.*]] = alloca ppc_fp128*, align 8 // CHECK18-NEXT: [[VLA1_ADDR:%.*]] = alloca ppc_fp128*, align 8
// CHECK18-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK18-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK18-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [8 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [8 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [8 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [8 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [8 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [8 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [8 x i64], align 8
// CHECK18-NEXT: store float* [[A]], float** [[A_ADDR]], align 8 // CHECK18-NEXT: store float* [[A]], float** [[A_ADDR]], align 8
// CHECK18-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8 // CHECK18-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8
// CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK18-NEXT: store ppc_fp128* [[VLA1]], ppc_fp128** [[VLA1_ADDR]], align 8 // CHECK18-NEXT: store ppc_fp128* [[VLA1]], ppc_fp128** [[VLA1_ADDR]], align 8
// CHECK18-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK18-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK18-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64 // CHECK18-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
Show All 15 Lines
// CHECK18-NEXT: [[TMP13:%.*]] = mul nuw i64 [[TMP3]], [[TMP5]] // CHECK18-NEXT: [[TMP13:%.*]] = mul nuw i64 [[TMP3]], [[TMP5]]
// CHECK18-NEXT: [[TMP14:%.*]] = mul nuw i64 [[TMP13]], 8 // CHECK18-NEXT: [[TMP14:%.*]] = mul nuw i64 [[TMP13]], 8
// CHECK18-NEXT: [[TMP15:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP15:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK18-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float**
// CHECK18-NEXT: store float* [[TMP8]], float** [[TMP16]], align 8 // CHECK18-NEXT: store float* [[TMP8]], float** [[TMP16]], align 8
// CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP17:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to float** // CHECK18-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to float**
// CHECK18-NEXT: store float* [[TMP8]], float** [[TMP18]], align 8 // CHECK18-NEXT: store float* [[TMP8]], float** [[TMP18]], align 8
// CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK18-NEXT: [[TMP19:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK18-NEXT: store i64 0, i64* [[TMP19]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK18-NEXT: [[TMP20:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK18-NEXT: store i8* null, i8** [[TMP20]], align 8 // CHECK18-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.St**
// CHECK18-NEXT: [[TMP21:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK18-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP21]], align 8
// CHECK18-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.St** // CHECK18-NEXT: [[TMP22:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK18-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP22]], align 8 // CHECK18-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.St**
// CHECK18-NEXT: [[TMP23:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK18-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP23]], align 8
// CHECK18-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to %struct.St** // CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK18-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP24]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP24]], align 8
// CHECK18-NEXT: [[TMP25:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK18-NEXT: [[TMP25:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: store i64 0, i64* [[TMP25]], align 8 // CHECK18-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i64*
// CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK18-NEXT: store i64 [[TMP1]], i64* [[TMP26]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP26]], align 8 // CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64* // CHECK18-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK18-NEXT: store i64 [[TMP1]], i64* [[TMP28]], align 8 // CHECK18-NEXT: store i64 [[TMP1]], i64* [[TMP28]], align 8
// CHECK18-NEXT: [[TMP29:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK18-NEXT: [[TMP29:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK18-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64* // CHECK18-NEXT: store i8* null, i8** [[TMP29]], align 8
// CHECK18-NEXT: store i64 [[TMP1]], i64* [[TMP30]], align 8 // CHECK18-NEXT: [[TMP30:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK18-NEXT: [[TMP31:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK18-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to ppc_fp128**
// CHECK18-NEXT: store i64 8, i64* [[TMP31]], align 8 // CHECK18-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP31]], align 8
// CHECK18-NEXT: [[TMP32:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK18-NEXT: [[TMP32:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK18-NEXT: store i8* null, i8** [[TMP32]], align 8 // CHECK18-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to ppc_fp128**
// CHECK18-NEXT: [[TMP33:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK18-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP33]], align 8
// CHECK18-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to ppc_fp128** // CHECK18-NEXT: [[TMP34:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK18-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP34]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP34]], align 8
// CHECK18-NEXT: [[TMP35:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK18-NEXT: [[TMP35:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK18-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to ppc_fp128** // CHECK18-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i64*
// CHECK18-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP36]], align 8 // CHECK18-NEXT: store i64 [[TMP3]], i64* [[TMP36]], align 8
// CHECK18-NEXT: [[TMP37:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK18-NEXT: [[TMP37:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK18-NEXT: store i64 0, i64* [[TMP37]], align 8 // CHECK18-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK18-NEXT: [[TMP38:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK18-NEXT: store i64 [[TMP3]], i64* [[TMP38]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP38]], align 8 // CHECK18-NEXT: [[TMP39:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK18-NEXT: [[TMP39:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK18-NEXT: store i8* null, i8** [[TMP39]], align 8
// CHECK18-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64* // CHECK18-NEXT: [[TMP40:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK18-NEXT: store i64 [[TMP3]], i64* [[TMP40]], align 8 // CHECK18-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i64*
// CHECK18-NEXT: [[TMP41:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP41]], align 8
// CHECK18-NEXT: [[TMP42:%.*]] = bitcast i8** [[TMP41]] to i64* // CHECK18-NEXT: [[TMP42:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK18-NEXT: store i64 [[TMP3]], i64* [[TMP42]], align 8 // CHECK18-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64*
// CHECK18-NEXT: [[TMP43:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP43]], align 8
// CHECK18-NEXT: store i64 8, i64* [[TMP43]], align 8 // CHECK18-NEXT: [[TMP44:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK18-NEXT: [[TMP44:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK18-NEXT: store i8* null, i8** [[TMP44]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP44]], align 8
// CHECK18-NEXT: [[TMP45:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK18-NEXT: [[TMP45:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK18-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i64* // CHECK18-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to double**
// CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP46]], align 8 // CHECK18-NEXT: store double* [[VLA]], double** [[TMP46]], align 8
// CHECK18-NEXT: [[TMP47:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK18-NEXT: [[TMP47:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK18-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to i64* // CHECK18-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to double**
// CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP48]], align 8 // CHECK18-NEXT: store double* [[VLA]], double** [[TMP48]], align 8
// CHECK18-NEXT: [[TMP49:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK18-NEXT: store i64 [[TMP14]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_sizes, i32 0, i32 6), align 8
// CHECK18-NEXT: store i64 8, i64* [[TMP49]], align 8 // CHECK18-NEXT: [[TMP49:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6
// CHECK18-NEXT: [[TMP50:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5 // CHECK18-NEXT: store i8* null, i8** [[TMP49]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP50]], align 8 // CHECK18-NEXT: [[TMP50:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7
// CHECK18-NEXT: [[TMP51:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK18-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i64*
// CHECK18-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to double** // CHECK18-NEXT: store i64 [[TMP12]], i64* [[TMP51]], align 8
// CHECK18-NEXT: store double* [[VLA]], double** [[TMP52]], align 8 // CHECK18-NEXT: [[TMP52:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK18-NEXT: [[TMP53:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6 // CHECK18-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to i64*
// CHECK18-NEXT: [[TMP54:%.*]] = bitcast i8** [[TMP53]] to double** // CHECK18-NEXT: store i64 [[TMP12]], i64* [[TMP53]], align 8
// CHECK18-NEXT: store double* [[VLA]], double** [[TMP54]], align 8 // CHECK18-NEXT: [[TMP54:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7
// CHECK18-NEXT: [[TMP55:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK18-NEXT: store i8* null, i8** [[TMP54]], align 8
// CHECK18-NEXT: store i64 [[TMP14]], i64* [[TMP55]], align 8 // CHECK18-NEXT: [[TMP55:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP56:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6 // CHECK18-NEXT: [[TMP56:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: store i8* null, i8** [[TMP56]], align 8 // CHECK18-NEXT: [[TMP57:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152.region_id, i32 8, i8** [[TMP55]], i8** [[TMP56]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK18-NEXT: [[TMP57:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7 // CHECK18-NEXT: [[TMP58:%.*]] = icmp ne i32 [[TMP57]], 0
// CHECK18-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to i64* // CHECK18-NEXT: br i1 [[TMP58]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18-NEXT: store i64 [[TMP12]], i64* [[TMP58]], align 8
// CHECK18-NEXT: [[TMP59:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK18-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i64*
// CHECK18-NEXT: store i64 [[TMP12]], i64* [[TMP60]], align 8
// CHECK18-NEXT: [[TMP61:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7
// CHECK18-NEXT: store i64 4, i64* [[TMP61]], align 8
// CHECK18-NEXT: [[TMP62:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7
// CHECK18-NEXT: store i8* null, i8** [[TMP62]], align 8
// CHECK18-NEXT: [[TMP63:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP64:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP65:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK18-NEXT: [[TMP66:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152.region_id, i32 8, i8** [[TMP63]], i8** [[TMP64]], i64* [[TMP65]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK18-NEXT: [[TMP67:%.*]] = icmp ne i32 [[TMP66]], 0
// CHECK18-NEXT: br i1 [[TMP67]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152(float* [[TMP8]], %struct.St* [[TMP9]], i64 [[TMP1]], ppc_fp128* [[TMP10]], i64 [[TMP3]], i64 [[TMP5]], double* [[VLA]], i64 [[TMP12]]) #[[ATTR4:[0-9]+]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152(float* [[TMP8]], %struct.St* [[TMP9]], i64 [[TMP1]], ppc_fp128* [[TMP10]], i64 [[TMP3]], i64 [[TMP5]], double* [[VLA]], i64 [[TMP12]]) #[[ATTR4:[0-9]+]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: [[TMP68:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK18-NEXT: [[TMP59:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP68]]) // CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP59]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152 // CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPg_l152
// CHECK18-SAME: (float* [[A:%.*]], %struct.St* [[S:%.*]], i64 [[VLA:%.*]], ppc_fp128* [[VLA1:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], double* nonnull align 8 dereferenceable(8) [[VLA26:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK18-SAME: (float* [[A:%.*]], %struct.St* [[S:%.*]], i64 [[VLA:%.*]], ppc_fp128* [[VLA1:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], double* nonnull align 8 dereferenceable(8) [[VLA26:%.*]], i64 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8 // CHECK18-NEXT: [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK18-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 8 // CHECK18-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 8
▲ Show 20 Lines • Show All 85 Lines • ▼ Show 20 Lines
// CHECK18-NEXT: [[VLA1_ADDR:%.*]] = alloca ppc_fp128*, align 8 // CHECK18-NEXT: [[VLA1_ADDR:%.*]] = alloca ppc_fp128*, align 8
// CHECK18-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 // CHECK18-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK18-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [10 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [10 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [10 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [10 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [10 x i8*], align 8 // CHECK18-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [10 x i8*], align 8
// CHECK18-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 8
// CHECK18-NEXT: store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8 // CHECK18-NEXT: store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
// CHECK18-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8 // CHECK18-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8
// CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK18-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK18-NEXT: store ppc_fp128* [[VLA1]], ppc_fp128** [[VLA1_ADDR]], align 8 // CHECK18-NEXT: store ppc_fp128* [[VLA1]], ppc_fp128** [[VLA1_ADDR]], align 8
// CHECK18-NEXT: [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8 // CHECK18-NEXT: [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
// CHECK18-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK18-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK18-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK18-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK18-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 28 Lines
// CHECK18-NEXT: [[TMP20:%.*]] = sub i64 [[TMP18]], [[TMP19]] // CHECK18-NEXT: [[TMP20:%.*]] = sub i64 [[TMP18]], [[TMP19]]
// CHECK18-NEXT: [[TMP21:%.*]] = sdiv exact i64 [[TMP20]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK18-NEXT: [[TMP21:%.*]] = sdiv exact i64 [[TMP20]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK18-NEXT: [[TMP22:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP22:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.St** // CHECK18-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.St**
// CHECK18-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP23]], align 8 // CHECK18-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP23]], align 8
// CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK18-NEXT: [[TMP24:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to %struct.St** // CHECK18-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to %struct.St**
// CHECK18-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP25]], align 8 // CHECK18-NEXT: store %struct.St* [[TMP9]], %struct.St** [[TMP25]], align 8
// CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK18-NEXT: [[TMP26:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK18-NEXT: store i64 0, i64* [[TMP26]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP26]], align 8
// CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK18-NEXT: [[TMP27:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK18-NEXT: store i8* null, i8** [[TMP27]], align 8 // CHECK18-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i64*
// CHECK18-NEXT: [[TMP28:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK18-NEXT: store i64 [[TMP1]], i64* [[TMP28]], align 8
// CHECK18-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i64* // CHECK18-NEXT: [[TMP29:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK18-NEXT: store i64 [[TMP1]], i64* [[TMP29]], align 8 // CHECK18-NEXT: [[TMP30:%.*]] = bitcast i8** [[TMP29]] to i64*
// CHECK18-NEXT: [[TMP30:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK18-NEXT: store i64 [[TMP1]], i64* [[TMP30]], align 8
// CHECK18-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to i64* // CHECK18-NEXT: [[TMP31:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK18-NEXT: store i64 [[TMP1]], i64* [[TMP31]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP31]], align 8
// CHECK18-NEXT: [[TMP32:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK18-NEXT: [[TMP32:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: store i64 8, i64* [[TMP32]], align 8 // CHECK18-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to ppc_fp128**
// CHECK18-NEXT: [[TMP33:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK18-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP33]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP33]], align 8 // CHECK18-NEXT: [[TMP34:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP34:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK18-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to ppc_fp128** // CHECK18-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to ppc_fp128**
// CHECK18-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP35]], align 8 // CHECK18-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP35]], align 8
// CHECK18-NEXT: [[TMP36:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK18-NEXT: [[TMP36:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK18-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to ppc_fp128** // CHECK18-NEXT: store i8* null, i8** [[TMP36]], align 8
// CHECK18-NEXT: store ppc_fp128* [[TMP10]], ppc_fp128** [[TMP37]], align 8 // CHECK18-NEXT: [[TMP37:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK18-NEXT: [[TMP38:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK18-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i64*
// CHECK18-NEXT: store i64 0, i64* [[TMP38]], align 8 // CHECK18-NEXT: store i64 [[TMP3]], i64* [[TMP38]], align 8
// CHECK18-NEXT: [[TMP39:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK18-NEXT: [[TMP39:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK18-NEXT: store i8* null, i8** [[TMP39]], align 8 // CHECK18-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64*
// CHECK18-NEXT: [[TMP40:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK18-NEXT: store i64 [[TMP3]], i64* [[TMP40]], align 8
// CHECK18-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i64* // CHECK18-NEXT: [[TMP41:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK18-NEXT: store i64 [[TMP3]], i64* [[TMP41]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP41]], align 8
// CHECK18-NEXT: [[TMP42:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK18-NEXT: [[TMP42:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK18-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64* // CHECK18-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i64*
// CHECK18-NEXT: store i64 [[TMP3]], i64* [[TMP43]], align 8 // CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP43]], align 8
// CHECK18-NEXT: [[TMP44:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK18-NEXT: [[TMP44:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK18-NEXT: store i64 8, i64* [[TMP44]], align 8 // CHECK18-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i64*
// CHECK18-NEXT: [[TMP45:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP45]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP45]], align 8 // CHECK18-NEXT: [[TMP46:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK18-NEXT: [[TMP46:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK18-NEXT: store i8* null, i8** [[TMP46]], align 8
// CHECK18-NEXT: [[TMP47:%.*]] = bitcast i8** [[TMP46]] to i64* // CHECK18-NEXT: [[TMP47:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP47]], align 8 // CHECK18-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to double**
// CHECK18-NEXT: [[TMP48:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK18-NEXT: store double* [[VLA]], double** [[TMP48]], align 8
// CHECK18-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i64* // CHECK18-NEXT: [[TMP49:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK18-NEXT: store i64 [[TMP5]], i64* [[TMP49]], align 8 // CHECK18-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to double**
// CHECK18-NEXT: [[TMP50:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK18-NEXT: store double* [[VLA]], double** [[TMP50]], align 8
// CHECK18-NEXT: store i64 8, i64* [[TMP50]], align 8 // CHECK18-NEXT: store i64 [[TMP14]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 5), align 8
// CHECK18-NEXT: [[TMP51:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK18-NEXT: [[TMP51:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK18-NEXT: store i8* null, i8** [[TMP51]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP51]], align 8
// CHECK18-NEXT: [[TMP52:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK18-NEXT: [[TMP52:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK18-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to double** // CHECK18-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to %struct.St**
// CHECK18-NEXT: store double* [[VLA]], double** [[TMP53]], align 8 // CHECK18-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP53]], align 8
// CHECK18-NEXT: [[TMP54:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK18-NEXT: [[TMP54:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK18-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to double** // CHECK18-NEXT: [[TMP55:%.*]] = bitcast i8** [[TMP54]] to i32**
// CHECK18-NEXT: store double* [[VLA]], double** [[TMP55]], align 8 // CHECK18-NEXT: store i32* [[A3]], i32** [[TMP55]], align 8
// CHECK18-NEXT: [[TMP56:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK18-NEXT: store i64 [[TMP21]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 6), align 8
// CHECK18-NEXT: store i64 [[TMP14]], i64* [[TMP56]], align 8 // CHECK18-NEXT: [[TMP56:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6
// CHECK18-NEXT: [[TMP57:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5 // CHECK18-NEXT: store i8* null, i8** [[TMP56]], align 8
// CHECK18-NEXT: store i8* null, i8** [[TMP57]], align 8 // CHECK18-NEXT: [[TMP57:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7
// CHECK18-NEXT: [[TMP58:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK18-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to %struct.St**
// CHECK18-NEXT: [[TMP59:%.*]] = bitcast i8** [[TMP58]] to %struct.St** // CHECK18-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP58]], align 8
// CHECK18-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP59]], align 8 // CHECK18-NEXT: [[TMP59:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK18-NEXT: [[TMP60:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6 // CHECK18-NEXT: [[TMP60:%.*]] = bitcast i8** [[TMP59]] to i32**
// CHECK18-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to i32** // CHECK18-NEXT: store i32* [[B2]], i32** [[TMP60]], align 8
// CHECK18-NEXT: store i32* [[A3]], i32** [[TMP61]], align 8 // CHECK18-NEXT: [[TMP61:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7
// CHECK18-NEXT: [[TMP62:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK18-NEXT: store i8* null, i8** [[TMP61]], align 8
// CHECK18-NEXT: store i64 [[TMP21]], i64* [[TMP62]], align 8 // CHECK18-NEXT: [[TMP62:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8
// CHECK18-NEXT: [[TMP63:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6 // CHECK18-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to %struct.St**
// CHECK18-NEXT: store i8* null, i8** [[TMP63]], align 8 // CHECK18-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP63]], align 8
// CHECK18-NEXT: [[TMP64:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7 // CHECK18-NEXT: [[TMP64:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 8
// CHECK18-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to %struct.St** // CHECK18-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32**
// CHECK18-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP65]], align 8 // CHECK18-NEXT: store i32* [[A3]], i32** [[TMP65]], align 8
// CHECK18-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7 // CHECK18-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 8
// CHECK18-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32** // CHECK18-NEXT: store i8* null, i8** [[TMP66]], align 8
// CHECK18-NEXT: store i32* [[B2]], i32** [[TMP67]], align 8 // CHECK18-NEXT: [[TMP67:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK18-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK18-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i64*
// CHECK18-NEXT: store i64 4, i64* [[TMP68]], align 8 // CHECK18-NEXT: store i64 [[TMP12]], i64* [[TMP68]], align 8
// CHECK18-NEXT: [[TMP69:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7 // CHECK18-NEXT: [[TMP69:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK18-NEXT: store i8* null, i8** [[TMP69]], align 8 // CHECK18-NEXT: [[TMP70:%.*]] = bitcast i8** [[TMP69]] to i64*
// CHECK18-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8 // CHECK18-NEXT: store i64 [[TMP12]], i64* [[TMP70]], align 8
// CHECK18-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to %struct.St** // CHECK18-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 9
// CHECK18-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP71]], align 8 // CHECK18-NEXT: store i8* null, i8** [[TMP71]], align 8
// CHECK18-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 8 // CHECK18-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP73:%.*]] = bitcast i8** [[TMP72]] to i32** // CHECK18-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: store i32* [[A3]], i32** [[TMP73]], align 8 // CHECK18-NEXT: [[TMP74:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144.region_id, i32 10, i8** [[TMP72]], i8** [[TMP73]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK18-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK18-NEXT: [[TMP75:%.*]] = icmp ne i32 [[TMP74]], 0
// CHECK18-NEXT: store i64 4, i64* [[TMP74]], align 8 // CHECK18-NEXT: br i1 [[TMP75]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18-NEXT: [[TMP75:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 8
// CHECK18-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK18-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK18-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i64*
// CHECK18-NEXT: store i64 [[TMP12]], i64* [[TMP77]], align 8
// CHECK18-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK18-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i64*
// CHECK18-NEXT: store i64 [[TMP12]], i64* [[TMP79]], align 8
// CHECK18-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK18-NEXT: store i64 4, i64* [[TMP80]], align 8
// CHECK18-NEXT: [[TMP81:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 9
// CHECK18-NEXT: store i8* null, i8** [[TMP81]], align 8
// CHECK18-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP83:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK18-NEXT: [[TMP84:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK18-NEXT: [[TMP85:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144.region_id, i32 10, i8** [[TMP82]], i8** [[TMP83]], i64* [[TMP84]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK18-NEXT: [[TMP86:%.*]] = icmp ne i32 [[TMP85]], 0
// CHECK18-NEXT: br i1 [[TMP86]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK18: omp_offload.failed: // CHECK18: omp_offload.failed:
// CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144(%struct.St* [[TMP9]], i64 [[TMP1]], ppc_fp128* [[TMP10]], i64 [[TMP3]], i64 [[TMP5]], double* [[VLA]], %struct.St* [[THIS1]], i64 [[TMP12]]) #[[ATTR4]] // CHECK18-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144(%struct.St* [[TMP9]], i64 [[TMP1]], ppc_fp128* [[TMP10]], i64 [[TMP3]], i64 [[TMP5]], double* [[VLA]], %struct.St* [[THIS1]], i64 [[TMP12]]) #[[ATTR4]]
// CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK18-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK18: omp_offload.cont: // CHECK18: omp_offload.cont:
// CHECK18-NEXT: [[TMP87:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 // CHECK18-NEXT: [[TMP76:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP87]]) // CHECK18-NEXT: call void @llvm.stackrestore(i8* [[TMP76]])
// CHECK18-NEXT: ret void // CHECK18-NEXT: ret void
// //
// //
// CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144 // CHECK18-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPg_l144
// CHECK18-SAME: (%struct.St* [[S:%.*]], i64 [[VLA:%.*]], ppc_fp128* [[VLA1:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], double* nonnull align 8 dereferenceable(8) [[VLA26:%.*]], %struct.St* [[THIS:%.*]], i64 [[N:%.*]]) #[[ATTR2]] { // CHECK18-SAME: (%struct.St* [[S:%.*]], i64 [[VLA:%.*]], ppc_fp128* [[VLA1:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], double* nonnull align 8 dereferenceable(8) [[VLA26:%.*]], %struct.St* [[THIS:%.*]], i64 [[N:%.*]]) #[[ATTR2]] {
// CHECK18-NEXT: entry: // CHECK18-NEXT: entry:
// CHECK18-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 8 // CHECK18-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK18-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 // CHECK18-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 106 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 4 // CHECK19-NEXT: [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 4
// CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [8 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [8 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [8 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [8 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [8 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [8 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [8 x i64], align 4
// CHECK19-NEXT: store float* [[A]], float** [[A_ADDR]], align 4 // CHECK19-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK19-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 4 // CHECK19-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 4
// CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK19-NEXT: store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 4 // CHECK19-NEXT: store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK19-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
Show All 12 Lines
// CHECK19-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 8 // CHECK19-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 8
// CHECK19-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64 // CHECK19-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64
// CHECK19-NEXT: [[TMP13:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP13:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to float** // CHECK19-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to float**
// CHECK19-NEXT: store float* [[TMP5]], float** [[TMP14]], align 4 // CHECK19-NEXT: store float* [[TMP5]], float** [[TMP14]], align 4
// CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK19-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float**
// CHECK19-NEXT: store float* [[TMP5]], float** [[TMP16]], align 4 // CHECK19-NEXT: store float* [[TMP5]], float** [[TMP16]], align 4
// CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK19-NEXT: store i64 0, i64* [[TMP17]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK19-NEXT: [[TMP18:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP18:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK19-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK19-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to %struct.St**
// CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK19-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP19]], align 4
// CHECK19-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.St** // CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK19-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP20]], align 4 // CHECK19-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.St**
// CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK19-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP21]], align 4
// CHECK19-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.St** // CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK19-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP22]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK19-NEXT: [[TMP23:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK19-NEXT: [[TMP23:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: store i64 0, i64* [[TMP23]], align 4 // CHECK19-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP24]], align 4 // CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK19-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK19-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK19-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4
// CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32* // CHECK19-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK19-NEXT: store i32 [[TMP0]], i32* [[TMP28]], align 4 // CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK19-NEXT: [[TMP29:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK19-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to x86_fp80**
// CHECK19-NEXT: store i64 4, i64* [[TMP29]], align 4 // CHECK19-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP29]], align 4
// CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK19-NEXT: store i8* null, i8** [[TMP30]], align 4 // CHECK19-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to x86_fp80**
// CHECK19-NEXT: [[TMP31:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK19-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP31]], align 4
// CHECK19-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to x86_fp80** // CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK19-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP32]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP32]], align 4
// CHECK19-NEXT: [[TMP33:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK19-NEXT: [[TMP33:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK19-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to x86_fp80** // CHECK19-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK19-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP34]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP34]], align 4
// CHECK19-NEXT: [[TMP35:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK19-NEXT: [[TMP35:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK19-NEXT: store i64 0, i64* [[TMP35]], align 4 // CHECK19-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK19-NEXT: [[TMP36:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP36]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP36]], align 4 // CHECK19-NEXT: [[TMP37:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK19-NEXT: [[TMP37:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK19-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK19-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK19-NEXT: [[TMP38:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP38]], align 4 // CHECK19-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK19-NEXT: [[TMP39:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK19-NEXT: store i32 [[TMP2]], i32* [[TMP39]], align 4
// CHECK19-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK19-NEXT: [[TMP40:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP40]], align 4 // CHECK19-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK19-NEXT: [[TMP41:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK19-NEXT: store i32 [[TMP2]], i32* [[TMP41]], align 4
// CHECK19-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK19-NEXT: [[TMP42:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK19-NEXT: [[TMP42:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK19-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK19-NEXT: [[TMP43:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK19-NEXT: [[TMP43:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK19-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK19-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to double**
// CHECK19-NEXT: store i32 [[TMP2]], i32* [[TMP44]], align 4 // CHECK19-NEXT: store double* [[VLA]], double** [[TMP44]], align 4
// CHECK19-NEXT: [[TMP45:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK19-NEXT: [[TMP45:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK19-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK19-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to double**
// CHECK19-NEXT: store i32 [[TMP2]], i32* [[TMP46]], align 4 // CHECK19-NEXT: store double* [[VLA]], double** [[TMP46]], align 4
// CHECK19-NEXT: [[TMP47:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK19-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_sizes, i32 0, i32 6), align 4
// CHECK19-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK19-NEXT: [[TMP47:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6
// CHECK19-NEXT: [[TMP48:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5 // CHECK19-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK19-NEXT: [[TMP48:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7
// CHECK19-NEXT: [[TMP49:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK19-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32*
// CHECK19-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to double** // CHECK19-NEXT: store i32 [[TMP9]], i32* [[TMP49]], align 4
// CHECK19-NEXT: store double* [[VLA]], double** [[TMP50]], align 4 // CHECK19-NEXT: [[TMP50:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK19-NEXT: [[TMP51:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6 // CHECK19-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32*
// CHECK19-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to double** // CHECK19-NEXT: store i32 [[TMP9]], i32* [[TMP51]], align 4
// CHECK19-NEXT: store double* [[VLA]], double** [[TMP52]], align 4 // CHECK19-NEXT: [[TMP52:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 7
// CHECK19-NEXT: [[TMP53:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK19-NEXT: store i8* null, i8** [[TMP52]], align 4
// CHECK19-NEXT: store i64 [[TMP12]], i64* [[TMP53]], align 4 // CHECK19-NEXT: [[TMP53:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP54:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6 // CHECK19-NEXT: [[TMP54:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK19-NEXT: [[TMP55:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152.region_id, i32 8, i8** [[TMP53]], i8** [[TMP54]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK19-NEXT: [[TMP55:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7 // CHECK19-NEXT: [[TMP56:%.*]] = icmp ne i32 [[TMP55]], 0
// CHECK19-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to i32* // CHECK19-NEXT: br i1 [[TMP56]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19-NEXT: store i32 [[TMP9]], i32* [[TMP56]], align 4
// CHECK19-NEXT: [[TMP57:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK19-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to i32*
// CHECK19-NEXT: store i32 [[TMP9]], i32* [[TMP58]], align 4
// CHECK19-NEXT: [[TMP59:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7
// CHECK19-NEXT: store i64 4, i64* [[TMP59]], align 4
// CHECK19-NEXT: [[TMP60:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 7
// CHECK19-NEXT: store i8* null, i8** [[TMP60]], align 4
// CHECK19-NEXT: [[TMP61:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP62:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP63:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK19-NEXT: [[TMP64:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152.region_id, i32 8, i8** [[TMP61]], i8** [[TMP62]], i64* [[TMP63]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK19-NEXT: [[TMP65:%.*]] = icmp ne i32 [[TMP64]], 0
// CHECK19-NEXT: br i1 [[TMP65]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152(float* [[TMP5]], %struct.St* [[TMP6]], i32 [[TMP0]], x86_fp80* [[TMP7]], i32 [[TMP1]], i32 [[TMP2]], double* [[VLA]], i32 [[TMP9]]) #[[ATTR4:[0-9]+]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152(float* [[TMP5]], %struct.St* [[TMP6]], i32 [[TMP0]], x86_fp80* [[TMP7]], i32 [[TMP1]], i32 [[TMP2]], double* [[VLA]], i32 [[TMP9]]) #[[ATTR4:[0-9]+]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: [[TMP66:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: [[TMP57:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP66]]) // CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP57]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152
// CHECK19-SAME: (float* [[A:%.*]], %struct.St* [[S:%.*]], i32 [[VLA:%.*]], x86_fp80* [[VLA1:%.*]], i32 [[VLA2:%.*]], i32 [[VLA4:%.*]], double* nonnull align 4 dereferenceable(8) [[VLA26:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK19-SAME: (float* [[A:%.*]], %struct.St* [[S:%.*]], i32 [[VLA:%.*]], x86_fp80* [[VLA1:%.*]], i32 [[VLA2:%.*]], i32 [[VLA4:%.*]], double* nonnull align 4 dereferenceable(8) [[VLA26:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4 // CHECK19-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK19-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 4 // CHECK19-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 4
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK19-NEXT: [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 4 // CHECK19-NEXT: [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 4
// CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK19-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [10 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [10 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [10 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [10 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [10 x i8*], align 4 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [10 x i8*], align 4
// CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK19-NEXT: store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4 // CHECK19-NEXT: store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
// CHECK19-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 4 // CHECK19-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 4
// CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK19-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK19-NEXT: store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 4 // CHECK19-NEXT: store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 4
// CHECK19-NEXT: [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4 // CHECK19-NEXT: [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
// CHECK19-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 25 Lines
// CHECK19-NEXT: [[TMP18:%.*]] = sub i64 [[TMP16]], [[TMP17]] // CHECK19-NEXT: [[TMP18:%.*]] = sub i64 [[TMP16]], [[TMP17]]
// CHECK19-NEXT: [[TMP19:%.*]] = sdiv exact i64 [[TMP18]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK19-NEXT: [[TMP19:%.*]] = sdiv exact i64 [[TMP18]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.St** // CHECK19-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.St**
// CHECK19-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP21]], align 4 // CHECK19-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP21]], align 4
// CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.St** // CHECK19-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.St**
// CHECK19-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP23]], align 4 // CHECK19-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP23]], align 4
// CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK19-NEXT: store i64 0, i64* [[TMP24]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK19-NEXT: store i8* null, i8** [[TMP25]], align 4 // CHECK19-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4
// CHECK19-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32* // CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK19-NEXT: store i32 [[TMP0]], i32* [[TMP27]], align 4 // CHECK19-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK19-NEXT: store i32 [[TMP0]], i32* [[TMP28]], align 4
// CHECK19-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK19-NEXT: [[TMP29:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK19-NEXT: store i32 [[TMP0]], i32* [[TMP29]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: store i64 4, i64* [[TMP30]], align 4 // CHECK19-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to x86_fp80**
// CHECK19-NEXT: [[TMP31:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK19-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP31]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP31]], align 4 // CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP32:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to x86_fp80** // CHECK19-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to x86_fp80**
// CHECK19-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP33]], align 4 // CHECK19-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP33]], align 4
// CHECK19-NEXT: [[TMP34:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP34:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK19-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to x86_fp80** // CHECK19-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK19-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP35]], align 4 // CHECK19-NEXT: [[TMP35:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK19-NEXT: [[TMP36:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK19-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK19-NEXT: store i64 0, i64* [[TMP36]], align 4 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP36]], align 4
// CHECK19-NEXT: [[TMP37:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK19-NEXT: [[TMP37:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK19-NEXT: store i8* null, i8** [[TMP37]], align 4 // CHECK19-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32*
// CHECK19-NEXT: [[TMP38:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP38]], align 4
// CHECK19-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32* // CHECK19-NEXT: [[TMP39:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP39]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK19-NEXT: [[TMP40:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK19-NEXT: [[TMP40:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK19-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32* // CHECK19-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK19-NEXT: store i32 [[TMP1]], i32* [[TMP41]], align 4 // CHECK19-NEXT: store i32 [[TMP2]], i32* [[TMP41]], align 4
// CHECK19-NEXT: [[TMP42:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK19-NEXT: [[TMP42:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK19-NEXT: store i64 4, i64* [[TMP42]], align 4 // CHECK19-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32*
// CHECK19-NEXT: [[TMP43:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK19-NEXT: store i32 [[TMP2]], i32* [[TMP43]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP43]], align 4 // CHECK19-NEXT: [[TMP44:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK19-NEXT: [[TMP44:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK19-NEXT: store i8* null, i8** [[TMP44]], align 4
// CHECK19-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32* // CHECK19-NEXT: [[TMP45:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK19-NEXT: store i32 [[TMP2]], i32* [[TMP45]], align 4 // CHECK19-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to double**
// CHECK19-NEXT: [[TMP46:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK19-NEXT: store double* [[VLA]], double** [[TMP46]], align 4
// CHECK19-NEXT: [[TMP47:%.*]] = bitcast i8** [[TMP46]] to i32* // CHECK19-NEXT: [[TMP47:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK19-NEXT: store i32 [[TMP2]], i32* [[TMP47]], align 4 // CHECK19-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to double**
// CHECK19-NEXT: [[TMP48:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK19-NEXT: store double* [[VLA]], double** [[TMP48]], align 4
// CHECK19-NEXT: store i64 4, i64* [[TMP48]], align 4 // CHECK19-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 5), align 4
// CHECK19-NEXT: [[TMP49:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK19-NEXT: [[TMP49:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK19-NEXT: store i8* null, i8** [[TMP49]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP49]], align 4
// CHECK19-NEXT: [[TMP50:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK19-NEXT: [[TMP50:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK19-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to double** // CHECK19-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to %struct.St**
// CHECK19-NEXT: store double* [[VLA]], double** [[TMP51]], align 4 // CHECK19-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP51]], align 4
// CHECK19-NEXT: [[TMP52:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK19-NEXT: [[TMP52:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK19-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to double** // CHECK19-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to i32**
// CHECK19-NEXT: store double* [[VLA]], double** [[TMP53]], align 4 // CHECK19-NEXT: store i32* [[A3]], i32** [[TMP53]], align 4
// CHECK19-NEXT: [[TMP54:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK19-NEXT: store i64 [[TMP19]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 6), align 4
// CHECK19-NEXT: store i64 [[TMP12]], i64* [[TMP54]], align 4 // CHECK19-NEXT: [[TMP54:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6
// CHECK19-NEXT: [[TMP55:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5 // CHECK19-NEXT: store i8* null, i8** [[TMP54]], align 4
// CHECK19-NEXT: store i8* null, i8** [[TMP55]], align 4 // CHECK19-NEXT: [[TMP55:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7
// CHECK19-NEXT: [[TMP56:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK19-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to %struct.St**
// CHECK19-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to %struct.St** // CHECK19-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP56]], align 4
// CHECK19-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP57]], align 4 // CHECK19-NEXT: [[TMP57:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK19-NEXT: [[TMP58:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6 // CHECK19-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to i32**
// CHECK19-NEXT: [[TMP59:%.*]] = bitcast i8** [[TMP58]] to i32** // CHECK19-NEXT: store i32* [[B2]], i32** [[TMP58]], align 4
// CHECK19-NEXT: store i32* [[A3]], i32** [[TMP59]], align 4 // CHECK19-NEXT: [[TMP59:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 7
// CHECK19-NEXT: [[TMP60:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK19-NEXT: store i8* null, i8** [[TMP59]], align 4
// CHECK19-NEXT: store i64 [[TMP19]], i64* [[TMP60]], align 4 // CHECK19-NEXT: [[TMP60:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8
// CHECK19-NEXT: [[TMP61:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6 // CHECK19-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to %struct.St**
// CHECK19-NEXT: store i8* null, i8** [[TMP61]], align 4 // CHECK19-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP61]], align 4
// CHECK19-NEXT: [[TMP62:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7 // CHECK19-NEXT: [[TMP62:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 8
// CHECK19-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to %struct.St** // CHECK19-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32**
// CHECK19-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP63]], align 4 // CHECK19-NEXT: store i32* [[A3]], i32** [[TMP63]], align 4
// CHECK19-NEXT: [[TMP64:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7 // CHECK19-NEXT: [[TMP64:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 8
// CHECK19-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32** // CHECK19-NEXT: store i8* null, i8** [[TMP64]], align 4
// CHECK19-NEXT: store i32* [[B2]], i32** [[TMP65]], align 4 // CHECK19-NEXT: [[TMP65:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK19-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK19-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK19-NEXT: store i64 4, i64* [[TMP66]], align 4 // CHECK19-NEXT: store i32 [[TMP9]], i32* [[TMP66]], align 4
// CHECK19-NEXT: [[TMP67:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 7 // CHECK19-NEXT: [[TMP67:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK19-NEXT: store i8* null, i8** [[TMP67]], align 4 // CHECK19-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i32*
// CHECK19-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8 // CHECK19-NEXT: store i32 [[TMP9]], i32* [[TMP68]], align 4
// CHECK19-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to %struct.St** // CHECK19-NEXT: [[TMP69:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 9
// CHECK19-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP69]], align 4 // CHECK19-NEXT: store i8* null, i8** [[TMP69]], align 4
// CHECK19-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 8 // CHECK19-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32** // CHECK19-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: store i32* [[A3]], i32** [[TMP71]], align 4 // CHECK19-NEXT: [[TMP72:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144.region_id, i32 10, i8** [[TMP70]], i8** [[TMP71]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK19-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK19-NEXT: [[TMP73:%.*]] = icmp ne i32 [[TMP72]], 0
// CHECK19-NEXT: store i64 4, i64* [[TMP72]], align 4 // CHECK19-NEXT: br i1 [[TMP73]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 8
// CHECK19-NEXT: store i8* null, i8** [[TMP73]], align 4
// CHECK19-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK19-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32*
// CHECK19-NEXT: store i32 [[TMP9]], i32* [[TMP75]], align 4
// CHECK19-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK19-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32*
// CHECK19-NEXT: store i32 [[TMP9]], i32* [[TMP77]], align 4
// CHECK19-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK19-NEXT: store i64 4, i64* [[TMP78]], align 4
// CHECK19-NEXT: [[TMP79:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 9
// CHECK19-NEXT: store i8* null, i8** [[TMP79]], align 4
// CHECK19-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP81:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK19-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK19-NEXT: [[TMP83:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144.region_id, i32 10, i8** [[TMP80]], i8** [[TMP81]], i64* [[TMP82]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK19-NEXT: [[TMP84:%.*]] = icmp ne i32 [[TMP83]], 0
// CHECK19-NEXT: br i1 [[TMP84]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK19: omp_offload.failed: // CHECK19: omp_offload.failed:
// CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144(%struct.St* [[TMP6]], i32 [[TMP0]], x86_fp80* [[TMP7]], i32 [[TMP1]], i32 [[TMP2]], double* [[VLA]], %struct.St* [[THIS1]], i32 [[TMP9]]) #[[ATTR4]] // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144(%struct.St* [[TMP6]], i32 [[TMP0]], x86_fp80* [[TMP7]], i32 [[TMP1]], i32 [[TMP2]], double* [[VLA]], %struct.St* [[THIS1]], i32 [[TMP9]]) #[[ATTR4]]
// CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK19: omp_offload.cont: // CHECK19: omp_offload.cont:
// CHECK19-NEXT: [[TMP85:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK19-NEXT: [[TMP74:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP85]]) // CHECK19-NEXT: call void @llvm.stackrestore(i8* [[TMP74]])
// CHECK19-NEXT: ret void // CHECK19-NEXT: ret void
// //
// //
// CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144
// CHECK19-SAME: (%struct.St* [[S:%.*]], i32 [[VLA:%.*]], x86_fp80* [[VLA1:%.*]], i32 [[VLA2:%.*]], i32 [[VLA4:%.*]], double* nonnull align 4 dereferenceable(8) [[VLA26:%.*]], %struct.St* [[THIS:%.*]], i32 [[N:%.*]]) #[[ATTR2]] { // CHECK19-SAME: (%struct.St* [[S:%.*]], i32 [[VLA:%.*]], x86_fp80* [[VLA1:%.*]], i32 [[VLA2:%.*]], i32 [[VLA4:%.*]], double* nonnull align 4 dereferenceable(8) [[VLA26:%.*]], %struct.St* [[THIS:%.*]], i32 [[N:%.*]]) #[[ATTR2]] {
// CHECK19-NEXT: entry: // CHECK19-NEXT: entry:
// CHECK19-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 4 // CHECK19-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 4
// CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK19-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 103 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 4 // CHECK20-NEXT: [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 4
// CHECK20-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK20-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK20-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [8 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [8 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [8 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [8 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [8 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [8 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [8 x i64], align 4
// CHECK20-NEXT: store float* [[A]], float** [[A_ADDR]], align 4 // CHECK20-NEXT: store float* [[A]], float** [[A_ADDR]], align 4
// CHECK20-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 4 // CHECK20-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 4
// CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK20-NEXT: store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 4 // CHECK20-NEXT: store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave() // CHECK20-NEXT: [[TMP3:%.*]] = call i8* @llvm.stacksave()
Show All 12 Lines
// CHECK20-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 8 // CHECK20-NEXT: [[TMP11:%.*]] = mul nuw i32 [[TMP10]], 8
// CHECK20-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64 // CHECK20-NEXT: [[TMP12:%.*]] = sext i32 [[TMP11]] to i64
// CHECK20-NEXT: [[TMP13:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP13:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to float** // CHECK20-NEXT: [[TMP14:%.*]] = bitcast i8** [[TMP13]] to float**
// CHECK20-NEXT: store float* [[TMP5]], float** [[TMP14]], align 4 // CHECK20-NEXT: store float* [[TMP5]], float** [[TMP14]], align 4
// CHECK20-NEXT: [[TMP15:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP15:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float** // CHECK20-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to float**
// CHECK20-NEXT: store float* [[TMP5]], float** [[TMP16]], align 4 // CHECK20-NEXT: store float* [[TMP5]], float** [[TMP16]], align 4
// CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK20-NEXT: [[TMP17:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK20-NEXT: store i64 0, i64* [[TMP17]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP17]], align 4
// CHECK20-NEXT: [[TMP18:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP18:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK20-NEXT: store i8* null, i8** [[TMP18]], align 4 // CHECK20-NEXT: [[TMP19:%.*]] = bitcast i8** [[TMP18]] to %struct.St**
// CHECK20-NEXT: [[TMP19:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK20-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP19]], align 4
// CHECK20-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.St** // CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK20-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP20]], align 4 // CHECK20-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.St**
// CHECK20-NEXT: [[TMP21:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK20-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP21]], align 4
// CHECK20-NEXT: [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.St** // CHECK20-NEXT: [[TMP22:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK20-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP22]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP22]], align 4
// CHECK20-NEXT: [[TMP23:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK20-NEXT: [[TMP23:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: store i64 0, i64* [[TMP23]], align 4 // CHECK20-NEXT: [[TMP24:%.*]] = bitcast i8** [[TMP23]] to i32*
// CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP0]], i32* [[TMP24]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP24]], align 4 // CHECK20-NEXT: [[TMP25:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP25:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32* // CHECK20-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK20-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4 // CHECK20-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4
// CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32* // CHECK20-NEXT: store i8* null, i8** [[TMP27]], align 4
// CHECK20-NEXT: store i32 [[TMP0]], i32* [[TMP28]], align 4 // CHECK20-NEXT: [[TMP28:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK20-NEXT: [[TMP29:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK20-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to x86_fp80**
// CHECK20-NEXT: store i64 4, i64* [[TMP29]], align 4 // CHECK20-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP29]], align 4
// CHECK20-NEXT: [[TMP30:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP30:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK20-NEXT: store i8* null, i8** [[TMP30]], align 4 // CHECK20-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to x86_fp80**
// CHECK20-NEXT: [[TMP31:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK20-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP31]], align 4
// CHECK20-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to x86_fp80** // CHECK20-NEXT: [[TMP32:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK20-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP32]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP32]], align 4
// CHECK20-NEXT: [[TMP33:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK20-NEXT: [[TMP33:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK20-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to x86_fp80** // CHECK20-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i32*
// CHECK20-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP34]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP34]], align 4
// CHECK20-NEXT: [[TMP35:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK20-NEXT: [[TMP35:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK20-NEXT: store i64 0, i64* [[TMP35]], align 4 // CHECK20-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK20-NEXT: [[TMP36:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP36]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP36]], align 4 // CHECK20-NEXT: [[TMP37:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK20-NEXT: [[TMP37:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK20-NEXT: store i8* null, i8** [[TMP37]], align 4
// CHECK20-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32* // CHECK20-NEXT: [[TMP38:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP38]], align 4 // CHECK20-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32*
// CHECK20-NEXT: [[TMP39:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK20-NEXT: store i32 [[TMP2]], i32* [[TMP39]], align 4
// CHECK20-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* // CHECK20-NEXT: [[TMP40:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP40]], align 4 // CHECK20-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK20-NEXT: [[TMP41:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK20-NEXT: store i32 [[TMP2]], i32* [[TMP41]], align 4
// CHECK20-NEXT: store i64 4, i64* [[TMP41]], align 4 // CHECK20-NEXT: [[TMP42:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK20-NEXT: [[TMP42:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK20-NEXT: store i8* null, i8** [[TMP42]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP42]], align 4
// CHECK20-NEXT: [[TMP43:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK20-NEXT: [[TMP43:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK20-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32* // CHECK20-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to double**
// CHECK20-NEXT: store i32 [[TMP2]], i32* [[TMP44]], align 4 // CHECK20-NEXT: store double* [[VLA]], double** [[TMP44]], align 4
// CHECK20-NEXT: [[TMP45:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK20-NEXT: [[TMP45:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK20-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to i32* // CHECK20-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to double**
// CHECK20-NEXT: store i32 [[TMP2]], i32* [[TMP46]], align 4 // CHECK20-NEXT: store double* [[VLA]], double** [[TMP46]], align 4
// CHECK20-NEXT: [[TMP47:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK20-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_sizes, i32 0, i32 6), align 4
// CHECK20-NEXT: store i64 4, i64* [[TMP47]], align 4 // CHECK20-NEXT: [[TMP47:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6
// CHECK20-NEXT: [[TMP48:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5 // CHECK20-NEXT: store i8* null, i8** [[TMP47]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP48]], align 4 // CHECK20-NEXT: [[TMP48:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7
// CHECK20-NEXT: [[TMP49:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK20-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32*
// CHECK20-NEXT: [[TMP50:%.*]] = bitcast i8** [[TMP49]] to double** // CHECK20-NEXT: store i32 [[TMP9]], i32* [[TMP49]], align 4
// CHECK20-NEXT: store double* [[VLA]], double** [[TMP50]], align 4 // CHECK20-NEXT: [[TMP50:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK20-NEXT: [[TMP51:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6 // CHECK20-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to i32*
// CHECK20-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to double** // CHECK20-NEXT: store i32 [[TMP9]], i32* [[TMP51]], align 4
// CHECK20-NEXT: store double* [[VLA]], double** [[TMP52]], align 4 // CHECK20-NEXT: [[TMP52:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 7
// CHECK20-NEXT: [[TMP53:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK20-NEXT: store i8* null, i8** [[TMP52]], align 4
// CHECK20-NEXT: store i64 [[TMP12]], i64* [[TMP53]], align 4 // CHECK20-NEXT: [[TMP53:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP54:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6 // CHECK20-NEXT: [[TMP54:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: store i8* null, i8** [[TMP54]], align 4 // CHECK20-NEXT: [[TMP55:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152.region_id, i32 8, i8** [[TMP53]], i8** [[TMP54]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK20-NEXT: [[TMP55:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7 // CHECK20-NEXT: [[TMP56:%.*]] = icmp ne i32 [[TMP55]], 0
// CHECK20-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to i32* // CHECK20-NEXT: br i1 [[TMP56]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20-NEXT: store i32 [[TMP9]], i32* [[TMP56]], align 4
// CHECK20-NEXT: [[TMP57:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK20-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to i32*
// CHECK20-NEXT: store i32 [[TMP9]], i32* [[TMP58]], align 4
// CHECK20-NEXT: [[TMP59:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7
// CHECK20-NEXT: store i64 4, i64* [[TMP59]], align 4
// CHECK20-NEXT: [[TMP60:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 7
// CHECK20-NEXT: store i8* null, i8** [[TMP60]], align 4
// CHECK20-NEXT: [[TMP61:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP62:%.*]] = getelementptr inbounds [8 x i8*], [8 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP63:%.*]] = getelementptr inbounds [8 x i64], [8 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK20-NEXT: [[TMP64:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152.region_id, i32 8, i8** [[TMP61]], i8** [[TMP62]], i64* [[TMP63]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK20-NEXT: [[TMP65:%.*]] = icmp ne i32 [[TMP64]], 0
// CHECK20-NEXT: br i1 [[TMP65]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152(float* [[TMP5]], %struct.St* [[TMP6]], i32 [[TMP0]], x86_fp80* [[TMP7]], i32 [[TMP1]], i32 [[TMP2]], double* [[VLA]], i32 [[TMP9]]) #[[ATTR4:[0-9]+]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152(float* [[TMP5]], %struct.St* [[TMP6]], i32 [[TMP0]], x86_fp80* [[TMP7]], i32 [[TMP1]], i32 [[TMP2]], double* [[VLA]], i32 [[TMP9]]) #[[ATTR4:[0-9]+]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: [[TMP66:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: [[TMP57:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP66]]) // CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP57]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152 // CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z10array_funcPfP2StiPe_l152
// CHECK20-SAME: (float* [[A:%.*]], %struct.St* [[S:%.*]], i32 [[VLA:%.*]], x86_fp80* [[VLA1:%.*]], i32 [[VLA2:%.*]], i32 [[VLA4:%.*]], double* nonnull align 4 dereferenceable(8) [[VLA26:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK20-SAME: (float* [[A:%.*]], %struct.St* [[S:%.*]], i32 [[VLA:%.*]], x86_fp80* [[VLA1:%.*]], i32 [[VLA2:%.*]], i32 [[VLA4:%.*]], double* nonnull align 4 dereferenceable(8) [[VLA26:%.*]], i32 [[N:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4 // CHECK20-NEXT: [[A_ADDR:%.*]] = alloca float*, align 4
// CHECK20-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 4 // CHECK20-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 4
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK20-NEXT: [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 4 // CHECK20-NEXT: [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 4
// CHECK20-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 // CHECK20-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4
// CHECK20-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [10 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [10 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [10 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [10 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [10 x i8*], align 4 // CHECK20-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [10 x i8*], align 4
// CHECK20-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [10 x i64], align 4
// CHECK20-NEXT: store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4 // CHECK20-NEXT: store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
// CHECK20-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 4 // CHECK20-NEXT: store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 4
// CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 // CHECK20-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK20-NEXT: store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 4 // CHECK20-NEXT: store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 4
// CHECK20-NEXT: [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4 // CHECK20-NEXT: [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
// CHECK20-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP1:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4 // CHECK20-NEXT: [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
Show All 25 Lines
// CHECK20-NEXT: [[TMP18:%.*]] = sub i64 [[TMP16]], [[TMP17]] // CHECK20-NEXT: [[TMP18:%.*]] = sub i64 [[TMP16]], [[TMP17]]
// CHECK20-NEXT: [[TMP19:%.*]] = sdiv exact i64 [[TMP18]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) // CHECK20-NEXT: [[TMP19:%.*]] = sdiv exact i64 [[TMP18]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP20:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.St** // CHECK20-NEXT: [[TMP21:%.*]] = bitcast i8** [[TMP20]] to %struct.St**
// CHECK20-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP21]], align 4 // CHECK20-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP21]], align 4
// CHECK20-NEXT: [[TMP22:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP22:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.St** // CHECK20-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.St**
// CHECK20-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP23]], align 4 // CHECK20-NEXT: store %struct.St* [[TMP6]], %struct.St** [[TMP23]], align 4
// CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 // CHECK20-NEXT: [[TMP24:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK20-NEXT: store i64 0, i64* [[TMP24]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP24]], align 4
// CHECK20-NEXT: [[TMP25:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK20-NEXT: [[TMP25:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK20-NEXT: store i8* null, i8** [[TMP25]], align 4 // CHECK20-NEXT: [[TMP26:%.*]] = bitcast i8** [[TMP25]] to i32*
// CHECK20-NEXT: [[TMP26:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP0]], i32* [[TMP26]], align 4
// CHECK20-NEXT: [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32* // CHECK20-NEXT: [[TMP27:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK20-NEXT: store i32 [[TMP0]], i32* [[TMP27]], align 4 // CHECK20-NEXT: [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i32*
// CHECK20-NEXT: [[TMP28:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK20-NEXT: store i32 [[TMP0]], i32* [[TMP28]], align 4
// CHECK20-NEXT: [[TMP29:%.*]] = bitcast i8** [[TMP28]] to i32* // CHECK20-NEXT: [[TMP29:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK20-NEXT: store i32 [[TMP0]], i32* [[TMP29]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP29]], align 4
// CHECK20-NEXT: [[TMP30:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 1 // CHECK20-NEXT: [[TMP30:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: store i64 4, i64* [[TMP30]], align 4 // CHECK20-NEXT: [[TMP31:%.*]] = bitcast i8** [[TMP30]] to x86_fp80**
// CHECK20-NEXT: [[TMP31:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK20-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP31]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP31]], align 4 // CHECK20-NEXT: [[TMP32:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP32:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to x86_fp80** // CHECK20-NEXT: [[TMP33:%.*]] = bitcast i8** [[TMP32]] to x86_fp80**
// CHECK20-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP33]], align 4 // CHECK20-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP33]], align 4
// CHECK20-NEXT: [[TMP34:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP34:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK20-NEXT: [[TMP35:%.*]] = bitcast i8** [[TMP34]] to x86_fp80** // CHECK20-NEXT: store i8* null, i8** [[TMP34]], align 4
// CHECK20-NEXT: store x86_fp80* [[TMP7]], x86_fp80** [[TMP35]], align 4 // CHECK20-NEXT: [[TMP35:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK20-NEXT: [[TMP36:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 // CHECK20-NEXT: [[TMP36:%.*]] = bitcast i8** [[TMP35]] to i32*
// CHECK20-NEXT: store i64 0, i64* [[TMP36]], align 4 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP36]], align 4
// CHECK20-NEXT: [[TMP37:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK20-NEXT: [[TMP37:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK20-NEXT: store i8* null, i8** [[TMP37]], align 4 // CHECK20-NEXT: [[TMP38:%.*]] = bitcast i8** [[TMP37]] to i32*
// CHECK20-NEXT: [[TMP38:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP38]], align 4
// CHECK20-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32* // CHECK20-NEXT: [[TMP39:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP39]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP39]], align 4
// CHECK20-NEXT: [[TMP40:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK20-NEXT: [[TMP40:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK20-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32* // CHECK20-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32*
// CHECK20-NEXT: store i32 [[TMP1]], i32* [[TMP41]], align 4 // CHECK20-NEXT: store i32 [[TMP2]], i32* [[TMP41]], align 4
// CHECK20-NEXT: [[TMP42:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 3 // CHECK20-NEXT: [[TMP42:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK20-NEXT: store i64 4, i64* [[TMP42]], align 4 // CHECK20-NEXT: [[TMP43:%.*]] = bitcast i8** [[TMP42]] to i32*
// CHECK20-NEXT: [[TMP43:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK20-NEXT: store i32 [[TMP2]], i32* [[TMP43]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP43]], align 4 // CHECK20-NEXT: [[TMP44:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK20-NEXT: [[TMP44:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK20-NEXT: store i8* null, i8** [[TMP44]], align 4
// CHECK20-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to i32* // CHECK20-NEXT: [[TMP45:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK20-NEXT: store i32 [[TMP2]], i32* [[TMP45]], align 4 // CHECK20-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to double**
// CHECK20-NEXT: [[TMP46:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK20-NEXT: store double* [[VLA]], double** [[TMP46]], align 4
// CHECK20-NEXT: [[TMP47:%.*]] = bitcast i8** [[TMP46]] to i32* // CHECK20-NEXT: [[TMP47:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK20-NEXT: store i32 [[TMP2]], i32* [[TMP47]], align 4 // CHECK20-NEXT: [[TMP48:%.*]] = bitcast i8** [[TMP47]] to double**
// CHECK20-NEXT: [[TMP48:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 4 // CHECK20-NEXT: store double* [[VLA]], double** [[TMP48]], align 4
// CHECK20-NEXT: store i64 4, i64* [[TMP48]], align 4 // CHECK20-NEXT: store i64 [[TMP12]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 5), align 4
// CHECK20-NEXT: [[TMP49:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 // CHECK20-NEXT: [[TMP49:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5
// CHECK20-NEXT: store i8* null, i8** [[TMP49]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP49]], align 4
// CHECK20-NEXT: [[TMP50:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK20-NEXT: [[TMP50:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK20-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to double** // CHECK20-NEXT: [[TMP51:%.*]] = bitcast i8** [[TMP50]] to %struct.St**
// CHECK20-NEXT: store double* [[VLA]], double** [[TMP51]], align 4 // CHECK20-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP51]], align 4
// CHECK20-NEXT: [[TMP52:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK20-NEXT: [[TMP52:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK20-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to double** // CHECK20-NEXT: [[TMP53:%.*]] = bitcast i8** [[TMP52]] to i32**
// CHECK20-NEXT: store double* [[VLA]], double** [[TMP53]], align 4 // CHECK20-NEXT: store i32* [[A3]], i32** [[TMP53]], align 4
// CHECK20-NEXT: [[TMP54:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 5 // CHECK20-NEXT: store i64 [[TMP19]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 6), align 4
// CHECK20-NEXT: store i64 [[TMP12]], i64* [[TMP54]], align 4 // CHECK20-NEXT: [[TMP54:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6
// CHECK20-NEXT: [[TMP55:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 5 // CHECK20-NEXT: store i8* null, i8** [[TMP54]], align 4
// CHECK20-NEXT: store i8* null, i8** [[TMP55]], align 4 // CHECK20-NEXT: [[TMP55:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7
// CHECK20-NEXT: [[TMP56:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK20-NEXT: [[TMP56:%.*]] = bitcast i8** [[TMP55]] to %struct.St**
// CHECK20-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to %struct.St** // CHECK20-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP56]], align 4
// CHECK20-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP57]], align 4 // CHECK20-NEXT: [[TMP57:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK20-NEXT: [[TMP58:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6 // CHECK20-NEXT: [[TMP58:%.*]] = bitcast i8** [[TMP57]] to i32**
// CHECK20-NEXT: [[TMP59:%.*]] = bitcast i8** [[TMP58]] to i32** // CHECK20-NEXT: store i32* [[B2]], i32** [[TMP58]], align 4
// CHECK20-NEXT: store i32* [[A3]], i32** [[TMP59]], align 4 // CHECK20-NEXT: [[TMP59:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 7
// CHECK20-NEXT: [[TMP60:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 6 // CHECK20-NEXT: store i8* null, i8** [[TMP59]], align 4
// CHECK20-NEXT: store i64 [[TMP19]], i64* [[TMP60]], align 4 // CHECK20-NEXT: [[TMP60:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8
// CHECK20-NEXT: [[TMP61:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 6 // CHECK20-NEXT: [[TMP61:%.*]] = bitcast i8** [[TMP60]] to %struct.St**
// CHECK20-NEXT: store i8* null, i8** [[TMP61]], align 4 // CHECK20-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP61]], align 4
// CHECK20-NEXT: [[TMP62:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7 // CHECK20-NEXT: [[TMP62:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 8
// CHECK20-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to %struct.St** // CHECK20-NEXT: [[TMP63:%.*]] = bitcast i8** [[TMP62]] to i32**
// CHECK20-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP63]], align 4 // CHECK20-NEXT: store i32* [[A3]], i32** [[TMP63]], align 4
// CHECK20-NEXT: [[TMP64:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7 // CHECK20-NEXT: [[TMP64:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 8
// CHECK20-NEXT: [[TMP65:%.*]] = bitcast i8** [[TMP64]] to i32** // CHECK20-NEXT: store i8* null, i8** [[TMP64]], align 4
// CHECK20-NEXT: store i32* [[B2]], i32** [[TMP65]], align 4 // CHECK20-NEXT: [[TMP65:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK20-NEXT: [[TMP66:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 7 // CHECK20-NEXT: [[TMP66:%.*]] = bitcast i8** [[TMP65]] to i32*
// CHECK20-NEXT: store i64 4, i64* [[TMP66]], align 4 // CHECK20-NEXT: store i32 [[TMP9]], i32* [[TMP66]], align 4
// CHECK20-NEXT: [[TMP67:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 7 // CHECK20-NEXT: [[TMP67:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK20-NEXT: store i8* null, i8** [[TMP67]], align 4 // CHECK20-NEXT: [[TMP68:%.*]] = bitcast i8** [[TMP67]] to i32*
// CHECK20-NEXT: [[TMP68:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8 // CHECK20-NEXT: store i32 [[TMP9]], i32* [[TMP68]], align 4
// CHECK20-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to %struct.St** // CHECK20-NEXT: [[TMP69:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 9
// CHECK20-NEXT: store %struct.St* [[THIS1]], %struct.St** [[TMP69]], align 4 // CHECK20-NEXT: store i8* null, i8** [[TMP69]], align 4
// CHECK20-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 8 // CHECK20-NEXT: [[TMP70:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP71:%.*]] = bitcast i8** [[TMP70]] to i32** // CHECK20-NEXT: [[TMP71:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: store i32* [[A3]], i32** [[TMP71]], align 4 // CHECK20-NEXT: [[TMP72:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144.region_id, i32 10, i8** [[TMP70]], i8** [[TMP71]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK20-NEXT: [[TMP72:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 8 // CHECK20-NEXT: [[TMP73:%.*]] = icmp ne i32 [[TMP72]], 0
// CHECK20-NEXT: store i64 4, i64* [[TMP72]], align 4 // CHECK20-NEXT: br i1 [[TMP73]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20-NEXT: [[TMP73:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 8
// CHECK20-NEXT: store i8* null, i8** [[TMP73]], align 4
// CHECK20-NEXT: [[TMP74:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK20-NEXT: [[TMP75:%.*]] = bitcast i8** [[TMP74]] to i32*
// CHECK20-NEXT: store i32 [[TMP9]], i32* [[TMP75]], align 4
// CHECK20-NEXT: [[TMP76:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK20-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32*
// CHECK20-NEXT: store i32 [[TMP9]], i32* [[TMP77]], align 4
// CHECK20-NEXT: [[TMP78:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 9
// CHECK20-NEXT: store i64 4, i64* [[TMP78]], align 4
// CHECK20-NEXT: [[TMP79:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 9
// CHECK20-NEXT: store i8* null, i8** [[TMP79]], align 4
// CHECK20-NEXT: [[TMP80:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP81:%.*]] = getelementptr inbounds [10 x i8*], [10 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK20-NEXT: [[TMP82:%.*]] = getelementptr inbounds [10 x i64], [10 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK20-NEXT: [[TMP83:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB1]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144.region_id, i32 10, i8** [[TMP80]], i8** [[TMP81]], i64* [[TMP82]], i64* getelementptr inbounds ([10 x i64], [10 x i64]* @.offload_maptypes.2, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK20-NEXT: [[TMP84:%.*]] = icmp ne i32 [[TMP83]], 0
// CHECK20-NEXT: br i1 [[TMP84]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK20: omp_offload.failed: // CHECK20: omp_offload.failed:
// CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144(%struct.St* [[TMP6]], i32 [[TMP0]], x86_fp80* [[TMP7]], i32 [[TMP1]], i32 [[TMP2]], double* [[VLA]], %struct.St* [[THIS1]], i32 [[TMP9]]) #[[ATTR4]] // CHECK20-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144(%struct.St* [[TMP6]], i32 [[TMP0]], x86_fp80* [[TMP7]], i32 [[TMP1]], i32 [[TMP2]], double* [[VLA]], %struct.St* [[THIS1]], i32 [[TMP9]]) #[[ATTR4]]
// CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK20-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK20: omp_offload.cont: // CHECK20: omp_offload.cont:
// CHECK20-NEXT: [[TMP85:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 // CHECK20-NEXT: [[TMP74:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4
// CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP85]]) // CHECK20-NEXT: call void @llvm.stackrestore(i8* [[TMP74]])
// CHECK20-NEXT: ret void // CHECK20-NEXT: ret void
// //
// //
// CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144 // CHECK20-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2St7St_funcEPS_iPe_l144
// CHECK20-SAME: (%struct.St* [[S:%.*]], i32 [[VLA:%.*]], x86_fp80* [[VLA1:%.*]], i32 [[VLA2:%.*]], i32 [[VLA4:%.*]], double* nonnull align 4 dereferenceable(8) [[VLA26:%.*]], %struct.St* [[THIS:%.*]], i32 [[N:%.*]]) #[[ATTR2]] { // CHECK20-SAME: (%struct.St* [[S:%.*]], i32 [[VLA:%.*]], x86_fp80* [[VLA1:%.*]], i32 [[VLA2:%.*]], i32 [[VLA4:%.*]], double* nonnull align 4 dereferenceable(8) [[VLA26:%.*]], %struct.St* [[THIS:%.*]], i32 [[N:%.*]]) #[[ATTR2]] {
// CHECK20-NEXT: entry: // CHECK20-NEXT: entry:
// CHECK20-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 4 // CHECK20-NEXT: [[S_ADDR:%.*]] = alloca %struct.St*, align 4
// CHECK20-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK20-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 96 Lines • Show Last 20 Lines